Introduction

The unhindered climate change and its consequences indicates how important it is to protect the groundwater resources for sustaining agricultural activities. The conservation of groundwater resources is an easy and economical way to manage rather than treating to remove unwanted substances from it. Therefore, the possibility of groundwater pollution should be addressed more and supported by raising awareness in terms of quality and quantity and performing vulnerability assessment over the aquifer system.

Recently, the evidence of frequent droughts at Southern coast of Türkiye and in Cyprus is thoroughly analyzed and elaborated at different levels, in which common consequence of droughts resulted in water deficit (Payab and Türker 2019; Zittis et al. 2022; Topçu et al. 2022; and Soylu Pekpostalci et al., 2023). Water shortages due to droughts are evident in Cyprus, and the issue is recently minimized by transporting water in suspended pipes through the Mediterranean Sea from Türkiye (Kayan et al. 2022). In order to obtain optimum benefit from this project and develop the sustainability of aquifers, it is required to conduct research on susceptibility and contamination risk of groundwater resources.

Even though numerical modelling or integration of numerical modelling with geospatial data analysis are useful tools for generating the susceptibility and pollution risk assessment of groundwater resources, the application of index-based techniques and geostatistical interpolation approaches can be preferable specially when sufficient quantitative data is missing (Kumar et al. 2015). In general, the index-based models of susceptibility measurement are carried out under the title “vulnerability index” and intrinsic aquifer vulnerability has been qualitatively measured by means of geographic information systems (GIS) based indices. In addition, the inherent hydro-geological properties of an aquifer that do not vary over time are taken into consideration to assess the vulnerability of an aquifer. Among several models, the most popular and frequently preferred GIS-based vulnerability index is the DRASTIC model. The model was initially suggested by Aller et al. (1987).

DRASTIC model is an overlay and index model that was developed to provide vulnerability ratings for aquifers by combining several hydro-geological layers. Hence, it is a model developed to reflect the influence of various hydro-geological parameters overlaying the pertinent groundwater resources. A rate coefficient and a weight were allocated to each critical parameter based on their significant effect on the quality and quantity of the resources.

In the pertinent literature, researchers (Ariffin et al. 2016; Arya et al. 2020; Nahin et al. 2010, Bera et al. 2022) have analyzed groundwater vulnerability by implementing various DRASTIC modeling techniques. Ariffin et al. (2016) implement the DRASTIC model in Malaysia and assessed the pollution of groundwater sources by agricultural activities. Arya et al. (2020) performed a pilot study with DRASTIC in south India and demarcated areas prone to nitrate pollution including fertilizer usage and livestock wastes. Nahin et al. (2020) used the DRASTIC index model to delineate the low, moderate, and high saline areas of the salinity-affected regions in the southwest coastal region of Bangladesh. The collaborative use of DRASTIC method with multiple decision techniques is also used to modify the DRASTIC method. Among them the AHP- modified DRASTIC method was observed to give successful results (Karan et al. 2018; Saranya and Saravan, 2020). Lately, Bera et al. (2022) have used Analytic Hierarchy Process (AHP) techniques and DRASTIC to evaluate the nitrate contamination levels at Totko River basin.

Numerous studies conducted in the last few years have addressed the inadequacies of DRASTIC model by significantly altering or modifying the weights and rates assigned for each parameter. These approaches were proposed to improve the effect of weights and rates on hydrogeological properties. One technique that has been widely used in the pertinent literature is AHP (Saravanan et al. 2023; Soyaslan 2020). It outperforms the standard DRASTIC method. Furthermore, in comparison to the traditional DRASTIC approach, better results were obtained in Torkashvand’s (2021) research when genetic algorithm (GA) optimization method and entropy were used to adjust the weights. In addition to changes in rates and weights, the use of land use/land cover (LULC) as the eighth parameter is also examined in several studies (Khan and Jhariya 2019; Shanmugamoorthy et al. 2023). The use of LULC parameter is essential since it directly reflects the anthropogenic interferences into the nature of the groundwater system. It has been proven that addition of LULC produces more useful results. Along with LULC, the use of groundwater nitrate content has also maintained successful outcomes in several studies (Voutchkova et al. 2021; Wei et al. 2021).

The primary goal in this research was to investigate the susceptibility of the groundwater resources of Yeşilköy Aquifer by implementing DRASTIC model and perform field analysis for assessing groundwater quality or deterioration due to agricultural activities. A representative groundwater quality distribution map is to be maintained by uing “Water Quality Index”, demonstrating the diffusion of contamination within the aquifer. It was also intended to change the weights and rates used in the DRASTIC method by using multi-criteria decision-making analysis. For this purpose, a new DRASTIC-AHP model is generated. Finally, the validity of the modified DRASTIC model was tested by determining the relation between the vulnerability index map and the calculated water quality index (WQI) data.

Methodology

Among the overlay and index methods, for the evaluation of contamination risks and susceptibility levels of aquifers, DRASTIC method is one of the most preferable and implemented methods. The model divides the aquifer into seven different parameters, each carrying different properties of the aquifer. The properties are the depth between the soil surface and the water table (D), recharge variations in the watershed (R), properties of the aquifer medium (A) and soil medium (S), the surface topography (T), vadose zone (I), and the hydraulic conductivity (C) of the aquifer. The digital maps of above-mentioned properties are generated and overlay on top of each other for generating resultant map that is numerically depicting the vulnerability of an aquifer to a possible pollution risks. The complete methodological process utilized to disclose the vulnerability map of the aquifer is shown in Fig. 1.

Fig. 1
figure 1

Flow chart displaying the roadmap of the developed model

Full size image

The DRASTIC index value (DIV) is the summation of the product of the weight of each property with its assigned rating. DIV is defined as:

$$\:DIV={D}_{r}{D}_{w}+{R}_{r}{R}_{w}+{A}_{r}{A}_{w}+{S}_{r}{S}_{w}+{T}_{r}{T}_{w}+{I}_{r}{I}_{w}+{C}_{r}{C}_{w}$$
(1)

where two important subscripts “r” and “w” are regarded as the ratings and weights, respectively. DIV is the sum of the product of the weight coefficients that the method has previously determined separately for each parameter, and the rate values assigned in line with the measurement findings or existing data subject to the parameters. In Eq. (1), “w” ranges between 1 and 5 (Table 1). A high w value indicates that the assigned property is important in pollution potential. The “r” value varies between 1 and 10. When the assigned value is closer to 10, it indicates high pollution potential of the considered DRASTIC component.

Although vulnerability categorization ranges are not provided by the DRASTIC technique, the researcher can solve the vulnerability index either by utilizing most preferable alternatives assigned for vulnerability assessment criteria in the literature or applying their field expertise and experience. In this study, the vulnerability index is categorized into four based on Angle (1995) as very high vulnerability (DIV ≥ 200), high vulnerability (141 ≤ DIV < 200), moderate vulnerability (101 ≤ DIV ≤ 140), and low vulnerability (1 ≤ DIV ≤ 100).

Table 1 Ratings and weightings assigned for DRASTIC model by Aller et al. (1987)
Full size table

Despite its simplicity and success, the DRASTIC vulnerability mapping technique has numerous flaws (Pacheco et al. 2015). Sometimes the results of DRASTIC model itself does not give satisfactory results, hence, new rate and weight factors can be proposed. Thus, multiple-criteria decision-making analysis (MCDA) models such as, the Analytic Hierarchy Process (AHP) (Saaty 1987) can be used to modify DRASTIC to improve the judgments with various criteria. Building a hierarchy based on the available problem is the first step of AHP. The second step entails giving each level of the hierarchy a nominal value and devising a pairwise comparison judgment matrix. The values of these matrices are obtained by transforming the judgments into numerical values according to binary comparison table (Table 2) while respecting the principle of reciprocity. Each class is then compared to every other class by giving it a relative rating between 1 and 9, for the purpose of comparison.

Table 2 Saaty’s scale of relative significance (Saaty 1987)
Full size table

Interpretation of pair-wise comparison matrix (B) comprises the components of each level and their weights as B1, B2, B3, …, Bn and α1, α2, α3, …, αn, respectively. The relative significance of βi and βj is represented by βij. The pair-wise comparison matrix of B1, B2, …, Bn as B = [βij] is then written as;

$$\begin{aligned}\:\text{B}=\left\{{{\upbeta\:}}_{\text{i}\text{j}}\right\}\text{n}\text{*}\text{n}& =\left(\begin{array}{ccc}1&\:{{\upbeta\:}}_{12}\cdots\:&\:{{\upbeta\:}}_{1\text{n}}\\\:?&\:\ddots\:&\:?\\\:{{\upbeta\:}}_{\text{n}1}&\:{{\upbeta\:}{\upalpha\:}}_{\text{n}2}\cdots\:&\:1\end{array}\right) \\ & \quad = \\ \left(\begin{array}{ccc}1&\:{{\upalpha\:}}_{1}/{{\upalpha\:}}_{2}\cdots\:&\:{{\upalpha\:}}_{1}/{{\upalpha\:}}_{\text{n}}\\\:{{\upalpha\:}}_{2}/{{\upalpha\:}}_{\text{n}}&\:1\dots\:&\:{{\upalpha\:}}_{1}/{{\upalpha\:}}_{2}\\\:{{\upalpha\:}}_{1}/{{\upalpha\:}}_{\text{n}}&\:{{\upalpha\:}}_{\text{n}}/{{\upalpha\:}}_{2}\cdots\:&\:1\end{array}\right)\end{aligned}$$
(2)

In this matrix, the element, βij = 1/βji and thus, when i = j, βij = 1. A matrix is normalized using Eq. (3)

$$\:{\:{{\upbeta\:}}^{{\prime\:}}}_{\text{i}\text{j}}=\frac{{{\upbeta\:}}_{\text{i}\text{j}}}{\sum\:_{1}^{\text{n}}{{\upbeta\:}}_{\text{i}\text{j}}}$$
(3)

Lastly, factor weights are calculated using Eq. (4) as:

$$\:{w}_{i}=\frac{1}{n}\sum\:_{1}^{n}{{\upbeta\:}{\prime\:}}_{\text{i}\text{j}}$$
(4)

The consistency ratio (CR) is used in AHP to check the matrix’s consistency. The ratio of the consistency index (CI) to an average random consistency index (RI) determines the magnitude of CR as shown in Eq. (5). On the other hand, Eq. (6) is used to calculate the magnitude of the consistency index.

$$\:{C}_{R}=\frac{CI}{{R}_{I}}$$
(5)
$$\:CI=\frac{{\lambda\:}_{max}-n}{n-1}$$
(6)

where n is the matrix’s order, and λmax is the principal eigenvalue. When the matrix’s order assigns a value of 1 and 2 average random consistency index (RI) becomes zero. When n is 3 RI becomes 0.5 and as n increases RI also increases up to 1.69.

The CR measures the consistency of pairwise comparisons and used in AHP to assess the consistency utilized to generate a matrix. To accept the estimated weights, the CR for a 7 × 7 matrix must be less than or equal to 0.1. Models with CR > 0.1 are automatically rejected. While modification of DRASTIC method with AHP model improves the quality of vulnerability analysis of an aquifer, spatial and temporal water quality parameters can be meaningful when they are matched and compared with the vulnerability index map. In this sense, the spatial Water quality index (WQI) distribution of an aquifer can express the in-depth water quality more understandably and universally and add value to the vulnerability index map. The execution of weighted arithmetic WQI method begins with the calculation of unit weights.

$$\:{W}_{n}=\frac{K}{{S}_{n}}$$
(7)

where \(\:{W}_{n}\) stands for the unit weight, \(\:{S}_{n}\) symbolize the standard allowable value, and the proportional constant is denoted by \(\:K\). After that, the quality rating can be calculated as.

$$\:{q}_{n\:}=100\left[\frac{{V}_{n}-{V}_{i}}{{S}_{n}-{V}_{i}}\right]$$
(8)

where \(\:{q}_{n\:}\:\)symbolize quality rating and \(\:{V}_{n}\) represents estimated value of nth parameters, and \(\:{V}_{i}\) represents ideal value. As the quality rating values are calculated, WQI is found according to the Eq. (9) (Bouslah et al. 2017).

$$\:WQI=\frac{\sum\:{q}_{n}{W}_{n}}{\sum\:{W}_{n}}$$
(9)

The calculated overall \(\:WQI\) is evaluated according to the classification proposed by Bouslah et al. (2017).

Study area

As a consequence of frequent drought occurrences in the northern part of Cyprus, surface water resources are inadequate and limited (Payab and Türker 2023). Consequently, all water-dependent industries are driven to use groundwater resources. There are three main aquifers mainly used as groundwater resource; Güzelyurt/Morphou aquifer, Kyrenia Range limestone aquifer and Yeşilköy aquifer. The former is the primary aquifer in Cyprus as it used to supply domestic and irrigation water for years (Gökcekuș et al., 2018; Payab et al. 2020; Kayan et al. 2021; Demir et al. 2022). The Kyrenia Range limestone aquifer consists of sub-aquifers and supplies high quality potable water that is only used for domestic purposes (Jamal and Türker 2015). Yeşilköy Aquifer has been utilized for agricultural and drinking purposes in Karpaz Peninsula, since the late 1960s. The recharge area of Yeşilköy aquifer is primarly used for agricultural activities especially for planting Colocasia esculenta and Solanum tuberosum. The long-term over-pumping from the aquifer consequently lowered the water table and give birth to water quality degradation. The location of Yeşilköy aquifer and its geological and hydrogeological maps are shown in Fig. 2. Yeşilköy aquifer occupies a surface area of approximately 5.46 km² and its catchment area is approximately 6.39 km² sloping towards southwest with a gradient of 3.5%. When the borehole logs of the region are examined, it was observed that there were five main units. These are marl, sandstone, calcarenite, secondary limestone and alluvial deposits (soil) from bottom to the top. The unit forming the aquifer is Pliocene-Late Quaternary sandstone, calcarenite and secondary limestone. Oligocene-Miocene aged marls represent the flysch, which is the impermeable lower limit of the aquifer. The semi-permeable unit above the aquifer consists of Holocene-aged alluvial deposits. From all these data, it is clear that the rocks in the region consist of clastic sedimentary rocks. As it is known, clastic sedimentary rocks are formed by the re-precipitation of grains as a result of weathering and abrasion. Since marls, one of the clastic sedimentary rocks in the study area, are always stored in a water environment, it is understood that the region has gone through a long-term transgression-regression cycle. Due to these cycles, there are karst structures in the region (Kulak 2006). Therefore, in some drillings performed in the field, the first unit to come after the alluvium in some wells is calcarenite, while in other wells the first unit to come after the alluvium is secondary limestone.

Fig. 2
figure 2

(a) The geological map of Yeşilköy Aquifer, (b) The hydrogeological map of Yeşilköy aquifer. Both maps include the ephemeral streams and the wells

Full size image

The rainfall varies yearly and the average rainfall for the last eight years is 376.7 mm/year. The minimum annual average rainfall was monitored in 2017 as 183.8 mm/year, whereas, maximum was observed in 2019 as 559.5 mm/year. Within the basin there are two ephemeral streams occasionally flowing during heavy rains. According to the survey carried out at the field, there are approximately 200 wells. Due to the consistent drop in water level, roughly 50 wells have been deepened in the last decade. Among 200 wells only 6 of them are under the management and administration of the local authorities. Almost all of the remaining wells are private and used for agricultural purposes. While potable water requirement is met by the inter-basin water transfer project from Türkiye, agricultural water demand is still covered by intensive pumping from Yeşilköy aquifer.

Results and discussions

Aquifer vulnerability analysis by DRASTIC model

To implement the DRASTIC technique, the weight of parameters and the corresponding rate values were determined from Table 1 and the vulnerability map was established with the help of geographic information system tools.

The travel period of pollutants and their access to groundwater depends on the depth of the water table (D) that represents the vertical distance between soil surface and water table. Deep groundwater level allows more filtration process for the possible pollutants, decreasing the vulnerability of the aquifer. Hence, the shorter the depth of the water table, higher the groundwater vulnerability. The water level data of 22 wells were used to interpolate the representative water depth map (Fig. 3a) by using Kriging interpolation technique and the data were re-classified by using the reclassification tool prepared in ArcGIS environment. Since the water table depth of all the monitored wells were more than 30 m, the rate factor for D is assigned as one (Fig. 4a).

Fig. 3
figure 3

The spatial distribution of (a) Depth, (b) Recharge rate, (c) Aquifer medium, (d) Soil media, (e) Topography, (f) Vadose zone, and, (g) conductivity maps

Full size image
Fig. 4
figure 4

The DRASTIC reclassified maps, (a) Depth, (b) Recharge rate, (c) Aquifer medium, (d) Soil media, (e) Topography, (f) Vadose zone, and, (g) conductivity maps and the resultant vulnerability maps of the Yeşilköy aquifer (h) Vulnerability Map based on DRASTIC method and (ı) DRASTIC vulnerability index map relative to Angle (1995)

Full size image

Recharge (R) of groundwater is significant in defining the vulnerability of groundwater. Together with recharging water, pollutants infiltrate through soil and can move vertically towards the water table. One of the hypothesis introduced in DRASTIC method is that the pollutant is transferred into the groundwater by rainfall (Aller et al. 1987 during the recharging process and it travels instantaneously with water. Due to high temperature rates, except in winter months, the evapotranspiration rate is higher than the precipitation rates. Hence, the net recharge to the aquifer from the irrigation is estimated to be 0.909 × 106 m³/year which is approximately 10% of the total water consumed for irrigation purposes (Türker et al., 2013). The water balance study performed by Türker et al. (2013) computed the annual volume of recharge due to precipitation as 1.01 × 106 m³/year over a recharging area of 6.39 km². Hence, the ratio of the total recharge volume to the recharge area per year results in the recharge rate of 173.7 mm/year as shown in Fig. 3b. Since 173.7 mm/year stands in between the range of 102–178 (Table 1), the net recharge rate was assigned as 6 and the weight of net recharge as 4 (Fig. 4b).

The size and the shape of the pores of the aquifer medium (A) impacts the vulnerability of the aquifer. It is well known that the concentration of pollutants is managed in part, by the aquifer’s capacity to reduce them. As the particle diameter of the materials building up the aquifer decreases, its permeability weakens and its vulnerability decreases. Yeşilköy aquifer is mostly built up with sandstone and calcarenite and contains clay, silt, marl, or their combined forms in varying amounts depending on location and depth. Assigned rate depending on the aquifer media content observed in each well varied between 5 and 8 (Fig. 4c). The inverse distance weighted approach was then used to interpolate the rates attributed to each well over the aquifer area. The resultant aquifer medium distribution rates map is given in Fig. 3c which depicts that the sand and gravel composition is more distributed at the southern part of the aquifer, whereas the calcarenite is dominant towards the north. Reclassified aquifer medium map is illustrated in Fig. 4c where sand and gravel are mostly dominant.

The texture of the soil layer medium (S) was accomplished by the field work performed at the site and adapting relevant laboratory testing analyses (Hydrometer Analyses). The soil samples were taken from close vicinity of 15 different wells, approximately 30 cm below the soil, by the help of auger. Sieve analyses were performed in the materials of construction laboratory of Eastern Mediterranean University. As a result of conducted sieve analyses, soil texture classification was performed for three different soil types. The results were grouped, and hydrometer analyses were performed as summarized in Table 3.

Table 3 Hydrometer analyses results of the samples
Full size table

Finally the soil texture chart is used to identify the texture of the soil samples. The texture was mainly consisting of loam and sandy loam (Fig. 3d). Accordingly, the rates for soil media are assigned as 5 and 6 (Fig. 4d).

The topography (T) is characterized by the slope of the region and it affects infiltration rates through the ground surface. The locations where steep slope dominates is incapable to retain water for long, which indicates less infiltration, and therefore the low amount of recharge. Topography also affects infiltration at ground surface. Gentle slope yields fewer runoff and further retention of water engendering more infiltration, and thus higher potential of contamination. Possible pollutants available over the soil surface may drain away if the surface gradient is steep enough. This will minimize the amount of pollutant percolating towards the aquifer. Therefore, T assigns the probability of a pollutant running off or percolating into the soil. In order to produce the topographical map, the 11-meter resolution Digital Elevation Model (DEM) map of Cyprus was used. The slope of the region varies between 0% and 76% as seen in Fig. 3e and the map is reclassified as illustrated in Fig. 4e. Reclassified map depicts that the upper surface of the Yeşilköy aquifer has a very high potential for water storage facilities.

The impact of the vadose zone (I) is the unsaturated region above the water table where pollutants travel before joining the groundwater. Hence, as the depth of vadose zone increase the risk of pollution decreases. In addition to vadose zone depth, the geological formations within the region are also essential as it can act as a possible barrier to decrease pollution by the process of adsorption and absorption subject upon the media (Karan et al. 2018). The sandstone, limestone, marl and calcarenite are the main formations that dominate the Yeşilköy aquifer’s vadose zone (Fig. 3f). In order to map the vadose zone 27 different boreholes were analyzed and studied. The rating values between 3 and 8 were recorded at varying locations (Fig. 4f). It was observed that, the north-eastern part of the aquifer is prone to be relatively less vulnerable than other locations.

The Hydraulic conductivity (C) is the property of an aquifer medium that describes the ability to transmit water along the pores of the fractures and the voids. Hence, it plays a key role in diffusion of pollutants within the aquifer medium. As the hydraulic conductivity increases, the risk of pollutant transmission within the aquifer increases. Based on Türker et al. (2013), the transmissivity of Yeşilköy aquifer is 1,782 m²/day. The transmissivity is the product of hydraulic conductivity and the thickness of an aquifer. Depending on the various thickness of Yeşilköy aquifer, the hydraulic conductivity varies between 0.03 and 0.29 m/day (Fig. 3g). According to Table 1, the rate value for conductivity ranging between 0 and 4 refers to a rate of 1 as shown in Fig. 4g.

According to the DRASTIC method, seven different hydro-geologic information-based maps are prepared, containing the weight and the rates of each parameter. This process is followed by matching seven maps to obtain the result of Eq. (11) representing the DRASTIC vulnerability map of Yeşilköy aquifer (Fig. 4h). The vulnerability index value have varied between 89 and 122. The vulnerability is lower at the North-eastern part of the aquifer (89 < DIV < 100), and increases towards the South-western part (100 < DIV < 122).

The vulnerability map can be reclassified into definite ranges so that regional vulnerability can be assigned for Yeşilköy Aquifer. In order to do so, the reclassified DRASTIC index map was created using the scale suggested by Angle (1995), (Fig. 4ı). Hence, 8.3% of the studied area can be categorized as “Low vulnerable”, and 91.7% of the studied area can be categorized as “Medium vulnerable”. Figure 4ı depicts that the vulnerability of the North-East region is less than other regions. The findings of this study has shown that, the Yeşilköy is vulnerable for environmental risks. On the other hand, there are environmental risks ongoing at the study area. These risks are fertilizers used during the agricultural activities, lack of sewerage system for the settlements in the area (buried septic tanks are active) and, solid waste dumping areas in and around the aquifer replenishing area. Hereafter, the impact of these pollution sources is to be analyzed by investigating the WQI of the aquifer.

Water quality analyses

The quality of water varies depending on the distribution of pollutants in the aquifer system. Hence, water quality distribution must be performed including a set of chemical, biological and physical parameters. Under these circumstances parameters like nitrite, nitrate, conductivity, mercury, pH, chloride, bromide, calcium, fluoride, sulphate, bicarbonate, sodium, potassium, magnesium, salinity (as NaCl), carbonate hardness (as CaCO3), permanent hardness (as CaCO3), total hardness, coliform, E. Coli, and Enterococcus Bacteria were examined to explore the contamination level and stresses in the Yeşilköy aquifer. Unexpectedly, coliform bacteria content were in large numbers, mainly over 2400. Mercury (Hg) was ranging in between 0.03 µg/L to 0.07 µg/L, which may be introduced into the aquifer by means of fertilizers used for agricultural activities. The pH value was 7.4 on the average and OH− ions were slightly higher. Conductivity content range from 905 µS/cm to 1990 µS/cm, none of which exceeds the 2500 µS/cm threshold. Indicating the prevalence of Sulphate in the wells, the critical threshold of 250 mg/L was determined to be surpassed at only one well (YK-17) with a value of 485 mg/L. The Sulphate varied between 8 mg/L and 485 mg/L, and the average value was 55.08 mg/L. Considering the total hardness, the average value was determined as 370 mg/L, varying between 0 mg/L and 754 mg/L. Chloride levels range from 101 mg/L to 405 mg/L, with an average of 192 mg/L. The magnesium content varies between 21 mg/L and 107 mg/L, and the average value was 45 mg/L. The bromide existence varies between 375 µg/L and 1595 µg/L, with an average of 674 µg/l. The presence of the calcium mineral is between 39 mg/L and 127 mg/L and the average value was 79.8 mg/L.

The WQI is a simple tool that is represented by a single value. It consists of four main steps which are; parameter selection, sub-indexing, parameter weighting and aggregating functions (Uddin et al. 2021). In the case of Yeşilköy aquifer, the WQI was calculated using 11 chemical and microbiological data including, Nitrate, Conductivity, pH, Chloride, Bromide, Sulphate, Sodium, Magnesium, Calcium, Total hardness, and Coliform. The tabular solution for WQI by the weighted arithmetic method is given in Table 4. Based on the tabular solution WQI is found to be equivalent to 73 and according to scale given by Bouslah et al. (2017) the range and quality status is referred to as “Poor Water Quality” and is classified as “C”.

Table 4 Water quality index calculation with weighted aritmetic method
Full size table

The WQI distribution along the aquifer is mapped and the classification according to the quality is depicted in Fig. 5. Those places where WQI is between 51 and 75 are classified as “poor water quality” status. The poor water quality is obviously due to the high Nitrate and Coliform content. Along with a section on the western side, the zone of “Poor water quality” accounts for 37% of the total area and runs from the northeast to the south of the aquifer. Water in this region can be used for irrigation and industrial purposes.

Fig. 5
figure 5

Water Quality Index distribution along the aquifer

Full size image

The remaining fractions are primarily categorized as “Very poor water quality” occupying 58% of the total area. In this region only irrigation practices can be performed. Furthermore, a very small section of the aquifer’s north-northeast region (5%) is designated as “Unsuitable for drinking water” and should be used after a proper treatment.

Matching the water quality index map and DRASTIC index map give birth to the opportunity to compare the WQI and DIV at each well, and define the association between the two parameters. Since the comparison of the two parameters indicate that as one quantity (WQI) increases or decreases the other quantity (DIV) experiences a unit increase or decrease simultaneously, it is assumed that the association between the two parameters is linear (Fig. 6). In this context, the association can be found by using Pearson correlation coefficient. As the coefficient approaches to zero, it means that there is a weak or no relationship between the two random quantities. On the other hand, a coefficient of + 1.0 or -1.0 indicates perfect positive or negative correlation, respectively (Türker and Valyrakis 2021). The results indicate that the Pearson correlation coefficient between the DIV and the WQI of the order of 0.2, represents very little correlation between the two quantities. Hence, an attempt to improve the weights and rates of the original DRASTIC method with AHP is worked out to improve the quality of correlation between DIV and WQI.

Fig. 6
figure 6

The relationship between WQI and DIV

Full size image

Analytic hierarchy process (AHP) decision-making analysis

Findings of the study has shown that the results of DRASTIC method does not associate with the existing conditions. Although the DRASTIC model assigns moderate vulnerability for the Yeşilköy aquifer, the heavy agricultural activities, and the lack of sewerage system contaminated the aquifer more than it was expected. Therefore, the existing vulnerability map must be developed to reflect the existing environmental conditions. Since the DRASTIC parameters proved to give successful results in many cases, instead of changing the parameters, it is preferred to create new rate and weight factors by introducing AHP to the weight and rate assignments. Pairwise comparison matrix, that ranks the priority of each parameter with respect to others is given in Table 5 and helps to assign new weight values for each parameter of DRASTIC model. For the ranges (sub-classes) of all parameters, the same procedure was used to generate a pairwise matrix. As anticipated, employing AHP enhanced the influence of the Vadose zone (I) and Aquifer media (A), while decreasing the effect of the water Depth (D) parameter. In terms of other parameters, no substantial changes were made. The results of the modified DRASTIC model are then utilized to bring about the new vulnerability index map of the Yeşilköy aquifer (Fig. 7(h)).

Fig. 7
figure 7

The reclassified thematic maps created by modified DRASTIC method by AHP and Vulnerability index map obtained by using AHP. (a) Aquifer depth, (b) Recharge rate, (c) Aquifer medium, (d) Soil media, (e) Topography, (f) Vadose zone, (g) conductivity and, (h) vulnerability index maps

Full size image
Table 5 Pairwise comparison matrix for AHP
Full size table

Figure 7 shows the reclassified thematic maps created by inserting the rate data for each parameter into the GIS environment. Thus, vulnerability index map is obtained as shown in the Fig. 7(h). The modified DRASTIC model built with AHP was provided as an alternative to the original DRASTIC method. Several chemical and microbiological water quality analyses were performed, and WQI were investigated based on samples taken from different wells. The association of the modified DRASTIC model established with AHP and the water quality index improved from 0.2 to 0.61 indicating moderate positive correlation between the WQI and DIV. This study has depicted that modifying the weight and rate of parameters has enhanced and strengthened the level of reliability of the modified DRASTIC model.

On the other hand, in terms of correlation analysis, it appears that nitrate and coliform are the most important parameters on the overall magnitude of WQI. The nitrate and coliform reveal that areas with high WQI are related to elevated nitrate and coliform concentrations observed in wells YK 17, YK 116, and YK 131. These wells are located close to the living spaces of the people in the region, and the elevated WQI is supposed to be due to the infiltration of wastewater from the septic tanks. This indicates that infrastructural modification in the study area is required to minimize the anthropogenic polluting activities directly from the septic tanks. Hence, the abandonment of septic tanks and integration of the sewer system into the study area is vital for improving the correlation coefficient between DIV and the WQI and is important for assessing the contamination risk in the study area. Taking such measures, and minimizing the pollution of the wells YK 17, YK 116, and YK 131 can help to improve the correlation between the DIV and WQI up to 0.74, indicating high correlation.

The DRASTIC model is built on the natural behavior of aquifers, excluding the positive and negative effects of artificial impacts. Therefore, the impact of artificial pollution by chemical or biological factors does not have to be associated with DIV retrieved from the DRASTIC model. However, the modifications to the DRASTIC model can help to detect the risks and improve the correlation between the DRASTIC parameters and the artificial pollution factors.

Conclusions

The present susceptibility of the Yeşilköy aquifer was evaluated and mapped using the modified DRASTIC model. Water quality index was tested and plotted on maps using both the DRASTIC approach and AHP modified DRASTIC approach. According to the results, 91.7% of the aquifer was “Medium Vulnerable”, the remaining 8.3% was “Low vulnerable”. The AHP-modified method revealed that the north-east zones of the aquifer are more vulnerable. In addition, “Poor Water Quality” was identified for the Yeşilköy aquifer as a result of chemical and microbiological analyses performed on samples obtained from 25 wells and WQI assessed throughout 11 parameters. A statistically significant relationship was found between the DRASTIC method modified by AHP and the WQI. As a result, following testimonies can be noted,

  1. 1.

    It is a correct approach to set the weights of the “Aquifer Media” and “Vadose Zone” higher than the others parameters.

  2. 2.

    The ‘Water Depth’ criterion was less essential since the water table was too deep and uniform.

  3. 3.

    It is the “Topography” parameter that has the least effect among the parameters.

The “DRASTIC Vulnerability Index Map” obtained by analyzing and adjusting numerous factors in the DRASTIC method and AHP approaches, is a map that can be assessed in terms of management to protect the Yeşilköy aquifer and minimize probable contamination.

Assessing the vulnerability of the Yeşilköy aquifer as well as defining the current status of the aquifer by examining various chemical and microbiological parameters were significantly helpful for future development of agricultural sector and water consumption. More frequent and regular chemical and microbiological analyses should be performed, and the status of the groundwater should be regularly monitored. Although groundwater vulnerability maps give users useful information, the major information source is tangible and up-to-date chemical and microbiological data.

In order to evaluate groundwater contamination risk from anthropogenic activities and determine a holistic strategy, vulnerability map of all aquifers throughout the island should be performed and pollution categorization network should be established. Such studies will assist regional planners and policymakers with sustainable ground water development and management opportunities.