Determination of groundwater potential distribution of Ceylanpinar Plain (Turkey) in Upper Mesopotamia by using geographical information techniques and Fuzzy-AHP with MCDM

Water is vital for life. Groundwater is a relatively abundant source of water in many parts of the world. Groundwater has a 30% share of fresh water resources worldwide (Ustun & Solmaz 2004; Ostad-Ali-Askari et al. 2019). Considering that approximately 70% of freshwater resources originate from polar glaciers and snow, the importance of efficient management of available groundwater is evident (Charbeneau 2000; Ostad-Ali-Askari et al. 2020). In Turkey, where the average precipitation is 643 mm, the water regime is irregular and the distribution of precipitation varies according to time and regions. When the data of many years is examined, the total usable water potential of Turkey is 110.0 billion m³/year, of which 98 billion m³/year is surface water and 12.0 billion m³/year is groundwater potential (World Water Council 2003). While the total water consumption in the world was 1,000 km³ in 1940, by 2040, worldwide water consumption is expected to exceed 4,350 billion m³ in terms of outflow (Statista 2021). Water consumption per capita in the world averages around 850 m³ per year. Considering that the world population is increasing by an average of 80 million people per year, it seems inevitable that the need for fresh water in the world will increase by 68 km³ per year (Sen 2003).

In comparison to surface waters, groundwater is the second most important freshwater reservoir and the best option for human and commercial activity (Ostad-Ali-Askari et al. 2017; Arabameri et al. 2019). Therefore, predicting and managing groundwater is crucial to avoid serious water shortages in drought-prone semi-arid regions (Deshmukh & Aher 2016; Golian et al. 2020). The geological and geomorphological structure of a region governs its hydrological environment and groundwater development (Doke et al. 2018; Das & Pal 2019; Ostad-Ali-Askari & Shayannejad 2021); where flow, seepage, percolation rates are controlled by physiographic factors. Some researchers consider relief, slope, drainage lithology, geological structures, soil, linear features, geomorphology, land use/land cover (LU/LC), rainfall amount, distance from the river, etc. It took into account that potential groundwater zones are generally determined by landscape, climatic, and environmental parameters (Das et al. 2017; Ostad-Ali-Askari et al. 2018a; Parameswari & Padmini 2018; Pande et al. 2020). River flow and environmental flow affect groundwater formation, hence natural streams and groundwater have a link (Murmu et al. 2019; Derakhshannia et al. 2020). The availability of groundwater fluctuates dramatically throughout the year, and river flows have a significant impact on the recharge process (Mohamed & Elmahdy 2017; Ostad-Ali-Askari et al. 2018b). The quality of groundwater is also affected by stream flow (Singh et al. 2018).

Groundwater exploration, for example, is a very specific decision-making task since it involves a variety of characteristics that are evaluated using competitive and unequal criteria (Malczewski 1999). The combination of these criteria aids in the development of a trustworthy and credible forecast map for future groundwater use planning in the region. Multi-criteria decision analysis (MCDA) can be used to achieve this integration. MCDA is an experienced and understood, math-based technique that allows you to choose between two or more competing options. It also aids decision makers in making the right decision from a range of alternatives and competing variables (Kan 2013). The most popular MCDA method is the analytical hierarchy process (AHP). AHP has today realized use in a variety of sectors of geology, most notably in groundwater research, where it has performed well, particularly in determining groundwater potential zones (Kan 2013; Fashae et al. 2014; Mogaji et al. 2014).

AHP is a biased method for examining control variables for comparing options. AHP uses binary comparisons to anticipate the relative importance of criteria and alternatives at all levels of hierarchy. AHP has been successfully coupled with other tools such as multi-objective mathematical programming, in addition to being used as a stand-alone decision tool (Ho 2008). Because traditional AHP is based on certainty and uncertainty, it is unable to accurately portray human thought (Kahraman et al. 2003). Van Laarhoven & Pedrycz (1983) suggested fuzzy analytical hierarchy process (FAHP) to overcome this problem and make a more confident decision. The FAHP methodology can be thought of as a more advanced version of regular AHP. The ease with which AHP can handle both quantitative and qualitative criteria in multi-criteria decision-making situations based on decision makers' judgments, as well as the uncertainty and incomprehensibility that exists in many decision-making situations, can all contribute to decision makers' ambiguous judgments in traditional AHP approaches (Bouyssou et al. 2000). In FAHP, criterion weights are considered to be predominantly determined by field experts using pairwise comparisons and subjective assessments to calculate weight ratios. Second, it is described as a set of triangular fuzzy numbers, the spread of which reflects the environment or lack of trust in a certain judgment. In practice, the process places a greater emphasis on the domain expert's more confident judgements. The parameters' scores can be either quantitative or qualitative. Subjective judgments of experts are used similarly to weights to create scores that are normalized to aspects that are naturally difficult to measure (Tan et al. 2013). Based on expert judgments, the FAHP approach turns the exact value into fuzzy integers and membership functions. This method allows for a more logical weighting of the criteria and, as a result, the best choices (Aryafar et al. 2013).

In this study, physical and mathematical analysis of the current state of the groundwater body of the Ceylanpinar Plain, which is one of Turkey's most significant plains with its 1.635.928 hectares of land and sustains most of its agricultural production with groundwater. The potential of the basin has been revealed by using numerical models through the evaluation of aquifer parameters and the analysis of pressure factors. In this context, studies on determining the geological and hydrogeological characteristics of the aquifer system of the basin, researching and defining the relationships, and calculating the hydrogeological parameters will be carried out.

The Ceylanpinar Plain groundwater system has been defined and necessary data and maps have been provided and analyzed within the scope of the study that will serve to use it efficiently and effectively for the benefit of humanity. The hydrometeorological module of the program, which will be used to calculate and model the current and future potential of the Ceylanpinar groundwater aquifer, has been prepared.

The Ceylanpinar Plain is one of the largest plains in Turkey and forms an important part of the upper Mesopotamian plain. The sustainable agricultural potential of the plain is proportional to the potential of groundwater. For this reason, this study is also important.

The study area, which is within the provincial borders of Şanlıurfa-Mardin, is between 360 40′ and 370 40′ north latitudes and 390 25′ and 400 40′ east longitudes according to Greenwich beginning. The average elevation of the plain is 500 m and the drainage area is 7,200 km². It is surrounded by ridges that form the topographic border with Karacadağ in the north, Tektek Mountains in the west, and the Kızıltepe Plain in the east. In this area, there is no other source other than Ceylanpinar source group, except for insignificant basalt covered springs and 0.5 m³/sec. Hanik spring is at the drainage border in the north. The average annual temperature in the region is 19 °C and the average precipitation is between 300 and 600 mm from south to north. The topographic subsidence is 6% in the north–south direction, and the altitude, which is 750 m on the skirts of Karacadağ, decreases to 350 m in the south around the springs. Grain is cultivated on a large scale in the region, which is very poor in terms of vegetation. These dry streams are seen as Eocene strips in the Miocene and they cover a distance of 50–70 km, starting from the foothills of Karacadağ and reaching the Turkey–Syria border, more or less parallel to each other. The Ceylanpinar Plain is located in southeastern Anatolia and represents a transboundary groundwater basin. The surface area of the plain is about 2,000 km². The Ceylanpinar Plain is the plain with the widest agricultural and animal use area in Turkey. The Ceylanpinar Plain is located in the lower Euphrates Basin in southeastern Anatolia region. Eocene-aged limestone forms the foundation of the Ceylanpinar Plain. These were covered by Miocene limestone, gravelly clay and clayey sands with gypsum in places. As we go north, basalt cover is located on these series. The most important formation that carries groundwater, the Eocene limestones, is located below the basalts in the north and the Miocene in the south and east. The Miocene formation usually consists of clay, limestone, marl, as well as a mixture of gravel and sand levels. Although the Miocene old formation carries a small amount of water, it is of poor quality. The thickness of the Miocene decreases as it moves from south to north and from west to east. Basalts spread from north to south, forming a cover partly on the Miocene series and partly on the Eocene limestones.

In dry agricultural areas where irrigation cannot be done, wheat and lentils are grown, while in watery areas wheat, cotton, lentils, sunflowers, feed plants are grown and corn is planted as the second crop. The Ceylanpinar Plain location map is presented in Figure 1.

In the equation, S represents the total groundwater suitability score, n represents the total number of groundwater suitability criteria, w_i represents the weight of the groundwater suitability criterion, and x_i represents the sub-criteria score of the groundwater suitability criterion.

In this study, land suitability assessments, which express the assignment of values to alternatives over more than one dimension or criteria, were made by using the MCDM method, which is used to solve real-world problems with multiple dimensions. In land suitability assessments, MCDM techniques, which include many tasks, including data collection, structuring, spatial analysis, and calculation of criteria, and the use of GIS (Joerin et al. 2001), are considered to be the most useful decision support systems (Mohit & Ali 2006). There are numerous approaches, each under the umbrella of MCDM, which include processes of interpreting and using formal results in the context of decision making, each using different protocols and many algorithms to reveal their inputs (Huang et al. 2011). AHP and its extensions analytic network process (ANP), simple additive weighting (SAW), technique for order of preference by similarity to ideal solution (TOPSIS), elimination et choix traduisant la realite (ELECTRE), grey relation analysis (GIA), the decision making trial and evaluation laboratory method (DEMATEL) Visekriterijumska Optimizacija I Kompromisno Resenje (VIKOR) and multi-objective optimization on the basis of ratio analysis (MOORA) are included in MCDM (Ersoy 2019). In this study, the use of AHP and FAHP, which are among the MCDM methods, was chosen as the method.

In recent years, the AHP approach for MCDM has been actively used and successfully implemented in groundwater potential zone mapping (Kaliraj et al. 2014). However, despite the demand for AHP, the method is sometimes criticized for failing to adequately deal with the inherent uncertainties and uncertainties associated with matching a decision maker's perception to sharp numbers (Chen et al. 2011). In dry and semi-arid environments, the AHP approach for MCDM has also been employed for groundwater potential mapping (Machiwal et al. 2011; Mallick et al. 2015; Rahmati et al. 2015). Due to the flexibility of fuzzy membership functions, none of the research combining fuzzy set theory with MCDA and specifically FAHP would increase the accuracy of groundwater potential maps. In this regard, the current study used a combined technique of GIS and FAHP to build thematic data layers for defining the GWPZ in Turkey's Ceylanpinar Basin. The following are the key objectives:

• Proposing and showing a methodology for identifying and delineating groundwater potential regions in the Ceylanpinar Basin utilizing GIS methods and FAHP through a case study;

• Recognize the variables that influence the identification of possible groundwater regions;

• It demonstrates the capabilities of GIS technology in groundwater mapping.

This study, which used the FAHP MCDA technique, will be useful for identifying possible groundwater regions that might assist decision makers, policymakers, and water resource managers in making integrated and reliable groundwater resource use decisions.

This research took into account a total of 10 different theme layers. The flow factor and water storage in the region are predicted to be controlled by these 10 layers. The influence of these variables was weighted based on their response to groundwater formation and expert judgment. A layer with a high weight suggests a layer with a strong impact on groundwater potential, while a layer with a low weight indicates a layer with a modest impact. Each criterion's weight was assigned according to its relative relevance rating on the Saaty scale (1–9). In addition, weights are assigned based on previous studies and field experience. The Saaty's relative importance value scale is evaluated between 9 and 1 according to its importance decency. According to the classification, weights were given according to the importance of the thematic layers and their water holding capacity. As a result, in the pairwise comparison matrix, all theme layers were compared to one another. The criteria questioned as to how important one is over the other are compared in pairs with respect to the next-level member, based on the judgment of the decision maker to determine their relative importance. In this way, a decision matrix is formed. This matrix is used to calculate the priorities of the criteria. In this evaluation, Saaty (1990) verbal evaluation scale corresponding to numerical values is used (Table 1).

For weight assignment, subclasses of thematic layers were classified in the GIS platform using the natural refractive classification method. Each thematic layer sequence's subclasses were ranked on a scale of 0–9 in terms of their relative impact on groundwater development. Table 2 shows the assigned order and weights of the thematic layers.

Saaty (1990) stated that a CR of 0.10 or less is acceptable to proceed with the analysis. If the CR value is 0, it means there is an excellent level of consistency in pairwise comparison.

In the FAHP method, comparisons are made in a matrix containing n elements. Since the comparison of a criterion with itself will be expressed as 1, all diagonal values of the matrix must be 1. After the completion of the binary comparisons, it is necessary to determine the relative importance of each element compared, i.e. its priority, which Saaty (1990) considers to be the ‘synthesis’ section. Saaty (1990) admitted that the priorities of the criteria can be estimated by finding the prime eigenvector of matrix A. Despite the fact that the AHP approach captures an expert's knowledge through perception or preference, it still falls short of properly reflecting human ideas with current numbers when compared to the FAHP method, due to the use of interval values rather than simple net numbers. To solve hierarchical fuzzy multi-criteria decision-making problems, the FAHP approach is used. The field experts' binary comparison of the decision factors is given using triangular fuzzy numbers in the FAHP approach.

The flow diagram for this study is given in Figure 2, and as shown in the figure, 10 hydrogeological parameters to evaluate the groundwater potential distribution, which are rainfall, soil class, land use, land cover, geology, geomorphology, hydrogeology, aquifer depth, slope, and drainage density maps were created. Table 1 shows the scale for pairwise comparison of one feature with the other in the FAHP method. Instead of using traditional numbers, the fuzzy weighting ratios of 2/3, 1, 3/2, 2, 5/2, 3, 7/2, 4, and 9/2 are utilized to represent the strength of preference of one element over another using interval numbers. The fuzzy scale was used to make pairwise comparisons of the parameters.

The processes of computing the relative weights of the criteria using FAHP were applied according to Chang (1996) measure analysis.

The values are needed to compare M₁ and M₂.

Third stage: The probability of the degree of convex fuzzy number being greater than k convex fuzzy number

In our study, we used the AHP and FAHP methodologies to give weight to each theme and its classes. Expert opinion, relevant literature, and field experience were used to generate comparison matrices as shown in Tables 4 and 5. Each subject in the comparison matrix was given equal weight in the comparison matrix. Precipitation and aquifers were given the most weight because they represent real groundwater recharging status. After that, there was soil class, slope, geology, geomorphology, drainage density, and land structure. The 10 thematic layers' consistency rates and the weights assigned to their features were in line with the expected outcomes. The slope played a significant part in the groundwater potential recharging area process in this semi-arid region. The most essential components of groundwater recharge for any region are precipitation, aquifer, and soil class. Statistical models based on AHP, FAHP, and GIS were used to determine relations between the site of borehole data and 10 groundwater potential parameters. Expert opinions, local hydrologist opinions and literature reviews were taken into account according to the selection of criteria. The GWPI was developed using a GIS-based groundwater potential mapping approach after theme harmonization of groundwater search criteria. In most environmental modeling applications, such as water resource assessments, the reliability of forecasting analytics has been a significant concern.

With the help of the results obtained from Tables 4 and 5, the results of the weighting (W) of the thematic layers applied to FAHP techniques and the normalized weights (NW) are given in Table 6.

According to the results obtained, the rational states and classified states of thematic maps are shown in Figures 4 and 5, respectively.

Multi-criteria decision system 10 parameters with multiple evaluation process: soil class, slope, land use, rainfall, aquifer criteria, geology, geomorphology, drainage density, land structure, depth and GWPI values were obtained. By classifying this value, the groundwater potential region (GWPZ) final map was obtained (Figure 6).

In this thematic map, the GWPI value ranges between 370 and 617. It explains the classification ranges according to GWPZ values and the total ratio by basin in Table 7. As seen in Figure 4, groundwater potential is at a good level in the center of the Ceylanpinar Basin and partially towards the south. It is mostly moderate in the northern parts.

The groundwater area map was validated using data from 19 observation water wells located inside the basin limits. The GWPZ map represents the groundwater study area map with water well locations. In Figure 6. The groundwater potential of almost all existing pumping wells for irrigation purposes are rated as very weak, poor, medium, good, or very good for the area. As a result of this classification, three of the 19 reference well data were partially discordant. Of these, 16 are compatible with a one-to-one rating. This situation is summarized in Table 8.

The Ceylanpinar Plain is part of the Mesopotamia, the most important region in history. It has been an area of agricultural activities since the beginning of civilization known in history. Agricultural lands in the region are very efficient and the irrigation water supply is groundwater. The sustainability of agriculture and livestock in the region depends on the sustainability of these resources. On a regional basis, it is possible to create an underground water potential map by modeling the impact rates of large-scale water potential on groundwater. A parametric approach using the techniques of Rs, GIS, and AHP can reduce time, labor, and costs to minimum levels and thus make faster decisions for productive management of water resources. Despite the limitations of the criteria, such analysis is a valuable and practical tool for areas and areas (especially developing states) for the determination of water solutions due to data shortages. The sustainability of agriculture and livestock in the region depends on the sustainability of these resources. On a regional basis, it is possible to create an underground water potential map by modeling the impact rates of large-scale water potential on groundwater.

Exploration of groundwater resources is one of the most important research fields in arid and semi-arid climates. Due to the agricultural expansion of the site, it was necessary to analyze the groundwater potential. In this study, a weighted thrust evaluation method employing GIS-based AHP and FAHP techniques was utilized to explore groundwater potential regions in the basin. The ability to employ fewer validation data (such as well locations and yield data) and the thematic layers of hydro geomorphology while still retaining optimal mapping accuracy is a benefit of such a mapping methodology. Totally, 10 thematic layers, such as geology, slope, aquifer, precipitation, depth, drainage density, land use, land structure, geomorphology, and soil texture were defined and classified as thematic layers by integrating and analyzing them in the GIS environment. Potential groundwater locations were then identified and classified into five categories: very good potential, good potential, medium potential, poor potential, and very poor potential. As a result, approximately 30% (472.19 km²) of the area has very bad potential; 23% (358.88 km²) has bad potential. 19% (298.27 km²): medium and 22% (350.9 km²): good groundwater potential suitable for agriculture. The resulting map categories align well with the productivity and sustainability of already developed groundwater resources and can therefore be used with some confidence to identify additional areas for development.

Groundwater well recharge, groundwater depth level data, and cascade drop pumping well tests should be performed in various sections of the basin to evaluate the particular efficiency of groundwater wells at various GWPIs, as described in this article. In this approach, the study site's groundwater can be extracted in a sustainable manner. However, in the absence of rigorous validation of groundwater data, the proposed methodology can be used as the primary estimation of groundwater expectations and the potential sites identified for drilling water wells should be preferred.

This study should contribute significantly to the understanding of the potential of groundwater zones in the studied basin, as well as the possibility of being used as a starting point by engineers, hydrologists, decision makers, and regional planners for the regeneration of this vital life-sustaining resource. Because the study method is based on logical criteria, it might be applied to various semi-arid and arid parts of the world with minor alterations. In order to build successful policies and processes for the healthy use and exploration of groundwater resources, a unified strategy is also required, particularly by ministries, government agencies, non-governmental organizations, and individuals. Future studies may aim to explore groundwater potential with fuzzy subjects, neural networks and other advanced artificial intelligence methods. Also, to better comprehend groundwater potential, groundwater criteria should be included as a decision-making approach.

All relevant data are included in the paper or its Supplementary Information.