Geographic information system and multi-criteria decision analysis for the determination of groundwater recharge potential: systematic review Open Access

The groundwater recharge area is one of the crucial components in regional development. This is because the groundwater recharge area functions as an area capable of infiltrating surface water or rainwater through pores or gaps in the soil or rock by gravity to the impermeable zone (Balek 1988; Zaidi et al. 2015; Yeh et al. 2016). So, it can be seen that in an area, groundwater recharge areas can reduce the risk of rising surface water causing flooding and maintain groundwater resources (Manna et al. 2017; da Costa et al. 2019). In the Presidential Regulation of the Republic of Indonesia No. 54 of 2008 concerning Regional Spatial Planning for Jakarta, Bogor, Tangerang, Bekasi, Puncak, and Cianjur, the water recharge areas are one of the criteria in planning priority protected areas. Groundwater recharge has two mechanisms, which are diffuse recharge by infiltration of atmospheric precipitation over a large area through the unsaturated zone to the aquifer and focused recharge by surface waterbodies such as rivers, streams, and lakes that lose water to aquifers or from water flow that moves through fractures (Ren et al. 2018; Reinecke 2022). In arid conditions, diffuse recharge becomes less important because of low precipitation and high evapotranspiration (Alley 2009; Ren et al. 2018).

Many studies in determining groundwater recharge areas have been carried out, such as by decision trees (Chenini & Ben Mammou 2010), weight-of-evidence (Lee et al. 2012) and frequency ratio (Ahmed et al. 2021) for those that are data-driven based, while for those that are physically based, one of them is the WetSpass model (Armanuos et al. 2016). Recent advances in digital image processing combined with conventional survey maps and multi-criteria analysis of the geospatial environment have allowed scientists to better identify areas. Natural groundwater recharge occurs using a combination of data such as geology, topography, land cover, soil type, lineament density, drainage density, slope, and rainfall (Lentswe & Molwalefhe 2020). The location selection procedure is carried out by combining and giving weight to geospatial data representing the research area according to the research objectives (Sallwey et al. 2019). The utilization of remote sensing and Geographic Information System (GIS) techniques can speed up and reduce costs in the qualitative determination of groundwater recharge areas (Adham et al. 2010). Multi-criteria analysis is a decision-making method to determine the best alternative from several alternatives that are evaluated based on specific criteria. Spatial analysis is generally used for the suitability of location selection (Malczewski 1999).

This paper review aims to determine the development of research on the application of GIS–Multi-Criteria Decision Analysis (GIS-MCDA) in mapping the potential for groundwater recharge. The analysis was carried out by knowing each location's dominant geological and climatic conditions, the criteria, decision-making methods, and the weighting used in obtaining the potential map in each case study, the weight of the criteria, and the validation process. A total of 31 case studies was obtained from the last ten years (2011–2021) on GIS-MCDA for determining the suitability of groundwater recharge using a systematic review and meta-analysis (PRISMA) procedure by taking English-based publications from the Scopus database. The development of research on the determination of groundwater recharge areas can be used as a lesson to practice in a sustainable spatial plan.

The limitation in conducting a review in selecting studies included in the inclusion criteria is that there are similarities, such as studies of artificial recharge and groundwater potential. To avoid these problems, after the abstract screening, full-text screening was carried out to determine the objectives of each study. With this process, we can obtain studies according to the inclusion criteria to be reviewed. In addition, the limitations in this systematic review limit the study sources to those only from journals so that other studies originating from conference papers are not included in the review process. The year used is only to review studies of the last ten years to determine the study's progress, so it is impossible to know the research condition before that year.

The following inclusion criteria were used as a guide in writing the review. The first inclusion is articles with the final publication stage, document types from articles, and sources from journals, in English. The second inclusion is the research aims to determine the suitability of the groundwater catchment area using the spatial multi-criteria method. As for the exclusion criteria in writing a review, these are the research aims to determine the suitability of artificial groundwater recharge areas, groundwater potential areas, and Managed Aquifer Recharge (MAR).

The final publication stage is used as a criterion because the search for articles begins on August 3, 2021, with a search year spanning 2011–2021. The types of documents selected in the criteria are articles, types of journal sources, and those in English. Journal articles are the source of most of the data included in systematic reviews (Li et al. 2021). The use of journal articles as a criterion is due to having complete information about methods and results. Meanwhile, English was chosen because it is the common language used by researchers in scientific publications. Inclusion 2 is used to get answers in the development of research on the application of GIS-MCDA for groundwater recharge, while exclusion criteria are used to avoid ambiguity in the data processing.

Scopus is the source of information used in the search for articles for writing a systematic review. Scopus is a multidisciplinary and selective database launched by Elsevier in November 2004 (Pranckutė 2021). Scopus has reproducible and documentable searches and has tools to analyze search results by author, affiliation, country, journal title, and broad subject categories (Gebre et al. 2021). Data was obtained on August 3, 2021. The search process uses the TITLE-ABS-KEY feature on the Scopus database with the string ((gis OR multi-criteria) AND (‘groundwater recharge’)) and published between 2011 and 2021 with document-type articles and sources from journals as well as in English and the final publication stage. Search results are exported and stored in MS Excel by listing the title, abstract, keywords, author and affiliation names, journal name, and year of publication.

The results of the search obtained are screened based on the eligibility criteria that have been made. Initial screening is carried out on abstracts to eliminate those that do not meet the criteria. The results of the screening are then carried out in full text or partial reading on articles that have not been eliminated from the previous stage to determine whether the article meets the eligibility criteria or falls into the exclusion criteria.

Data collection is carried out from the export of Scopus database information sources using MS Excel based on the name of the journal, year, and country, and added manually from the selected articles in the form of dominant geological information in the research area, the climate of the research area, research criteria, research criteria weights, decision techniques, and the process of verification or validation of results.

Information taken from each article consists of (1) information on the geology and climate of the research area, and (2) the criteria used in the research, the weight of the criteria in the research, decision-making techniques, and the process of verifying or validating research results.

The information from number 1 is to determine what kind of dominant research was carried out on geological and climatic conditions, while the information from number 2 is to find out the development and number of criteria used during the last ten years, the weight of the criteria used from all research and based on the grouping of decision-making techniques, decision-making techniques used during the last ten years, and how the results validation process is carried out.

Based on the search results from the Scopus database, 331 articles were obtained in 2011–2021 (August 3, 2021). After screening abstracts based on eligibility criteria, 95 articles were obtained; 31 articles were obtained after full-text and partial reading of articles based on eligibility and exclusion criteria. The 31 selected articles were used in the review.

Based on the dominant geology of each study, they are grouped into five classes, namely hard rock, soft rock, various deposits, volcanic deposits, unconsolidated deposits. Hard rock is a term for naming igneous and metamorphic rock types such as granite and gneiss, while soft rock names sedimentary rock types such as limestone and sandstone (Hall 2006). Volcanic deposits are the result of extrusive volcanic activity such as lava, and unconsolidated deposits are alluvial or eolian deposits. From the 31 studies, eight studies were obtained with dominant geology of hard rock and unconsolidated deposits, six studies with dominant geology of various deposits and soft rock, and three studies with dominant geology of volcanic deposits. Six of the eight studies with hard-rock-dominant geology are in India, Tanzania and Botswana. As for studies with dominant unconsolidated deposit geology, three are in India, two are in Saudi Arabia, and one is in New Zealand, Algeria, and Morocco. Studies with dominant geology of various deposits are in India for as many as three studies, and one study each in Taiwan, Egypt, and Nigeria. Studies with soft-rock dominant geology are in Taiwan, Tunisia, Iraq, Algeria, and Australia, with one study each. Studies with the dominant geology of volcanic deposits are in India with as many as two studies and one study is in Jordan. From all the geology or lithology types, the hard rock geological condition becomes the most-used in studies as a focus to determine groundwater recharge (Deepa et al. 2016; Kirubakaran et al. 2016; Fagbohun 2018; Patil et al. 2018; Mishra et al. 2020; Mussa et al. 2020; Fauzia et al. 2021).

Climate conditions were analysed using the Koppen Climate Classification, and the metadata for global climate data was obtained from Beck et al. (2018). Based on the Koppen Climate Classification and country distribution (shown in Figure 3), studies were done in Aw (tropical, savannah), all arid conditions (BWh, BWk, BSh, BSk), Csa (temperate, dry summer, hot summer), Cfa (temperate, no dry season, hot summer), Cfb (temperate, no dry season, warm summer), Cwa (temperate, dry summer, hot summer), Dfc (cold, no dry season, cold summer), and ET (polar, tundra) climate conditions. India has conducted studies with climate variations, namely BSh, Aw, and Cwa; and the same with New Zealand, and since the study was done in one country, the climate variations are Cfb, BSk, Dfc, and ET. The African continent and Western Asia mostly have semi-arid (BSh, BSk) and arid (BWh, BWk) climate conditions, one study with tropical (Aw) condition was done in Nigeria (Fagbohun 2018). In Australia, the study was only conducted under semi-arid (BSh-BSk) climate condition (Ahmed et al. 2021). Based on the climate conditions, arid and semi-arid climate conditions are usually used as a focus of study to determine groundwater recharge (Mahmoud et al. 2014; Ait El Mekki & Laftouhi 2016; Saidi et al. 2017; Al-Shabeeb 2018; Kadam et al. 2020; Lentswe & Molwalefhe 2020; Mussa et al. 2020; Zghibi et al. 2020).

Validation is carried out after getting the groundwater recharge potential map results. Generally, this is done by verifying well data in the research area. A total of 15 studies verified the groundwater recharge potential map results with well data. Three studies used water level fluctuation data as verification data for the map of potential groundwater recharge. Several studies have used comparisons of the results of groundwater recharge potential maps with maps or other spatial data such as downstream maps and agricultural land (Abdelkareem & Al-Arifi 2021), electrical conductivity data (Panda et al. 2020), existing dam data (Salar et al. 2018), geological maps, and watershed flow data to obtain a quick flow index (Singh et al. 2019), and data on existing groundwater infiltration structures (Mahmoud 2014). Seven studies did not validate or verify the groundwater recharge potential map results. ROC analysis was carried out in two studies using well data to compare accuracy for each method. The results found that the AHP method out performs the MIF method (Zghibi et al. 2020) and the FR method out performs the AHP and MIF methods (Ahmed et al. 2021).

Sensitivity analysis was carried out in five studies from 31 existing studies. Sensitivity analysis is a method to determine the sensitivity of each criterion by removing one or more thematic maps (MRSA) (Lodwick et al. 1990) and to understand the contribution of each criterion to the map results with a Single Parameter Sensitivity Analysis (SPSA) (Napolitano & Fabbri 1996). Map removal analysis to determine the value of sensitivity analysis was carried out by Al-Shabeeb (2018), while the study from Jena et al. (2020) used the percentage change in the area of the potential zone to the thematic layer. Ait El Mekki & Laftouhi (2016) and Boufekane et al. (2020) used the effective weight obtained by multiplying each criterion's rating value and weight value divided by the potential groundwater index. Meanwhile, Zghibi et al. (2020) used two types of sensitivity analysis, namely map removal and single parameter or effective weight.

Based on the systematic review results, a total of 31 studies sourced from journals on GIS-MCDA to determine potential groundwater recharge areas were carried out over the last ten years (2011–2021). India is the most dominant country in conducting studies to determine the potential for groundwater recharge. The dominant study of geology and climatic conditions of each area has variations. Some studies use geological conditions such as hard rock and arid and semi-arid climatic conditions as the study's focus. This is because the geological and climatic conditions are closely related to the presence and source of groundwater. Hard rock geological conditions prevent water from infiltrating to the aquifer because of low to impermeable conditions, groundwater recharge in hard rock depending on rocks that have weathered and secondary porosities, such as joints and fractures (Kirubakaran et al. 2016; Fauzia et al. 2021), while the conditions of arid and semi-arid climate are related to a lack of annual rainfall and high annual temperature (Ait El Mekki & Laftouhi 2016; Ahmed et al. 2021).

The development of criteria use over the last ten years has reached 33 variations of criteria. Of the 33 criteria, those generally used in each study are slope, LULC, lithology, drainage density, lineament density, soil, and rainfall. The seven criteria influence the value of the potential for groundwater recharge. In addition, these criteria have data availability and can be obtained through remote sensing and integrated with GIS and multi-criteria methods (Ahmed et al. 2021). In the last ten years from 2011 to 2021, there are about 18 new criteria that have been used only in one study, with the most recent study offering the integration of criteria from a conventional map with remote sensing, geophysical and borewell data (Fauzia et al. 2021). AHP and MIF have been commonly used for decision rules and weighting methods in the last ten years. Each method has advantages in determining the potential of groundwater catchment areas. The AHP method has an advantage because in determining the weights, it is determined by expert opinion and the consistency ratio for assessing the consistency of weights is <10% (Lentswe & Molwalefhe 2020). Meanwhile, MIF has the advantage of weighting based on the relationship between criteria (major or minor) (Huang et al. 2013). Frequency Ratio (FR) as the newest data-driven method in MCDA uses the ratio between the number of observational water wells and the criteria that influence groundwater recharge (Ahmed et al. 2021). FR result accuracy is shown to out-perform the AHP and MIF methods. However this method requires data availability to be performed.

The highest weight based on 31 studies on the seven criteria evaluated was the lithological criteria. Variations in the weight of the evaluation results occur due to differences in the location of the study, the method used, and the number of criteria used in the study. This affects the amount of weight in each criterion used in the study. The groundwater recharge potential map validation results based on MCDA have variations for the last ten years. The development of validation supports determining the accuracy of the groundwater recharge potential map results. However, only five out of 31 studies carried out sensitivity analysis and two studies carried out ROC analysis.

Variations in the geological and climatic conditions of each study and the number of criteria used lead to different assessments of each criterion. The AHP method considers weights based on expert opinion in each study country with different geological and climatic conditions that can cause differences in decision-making for the weight of the criteria. The MIF method considers weights based on the relationship between each criterion. In the weighting process, the difference in the number of criteria will affect the weight results of the criteria.

The systematic review findings on practice imply that the results collected can be used to determine the process in determining the potential for groundwater recharge areas, especially with the multi-criteria GIS method. The development of MCDA and GIS for groundwater recharge studies from 2011 to 2021 show that 12 countries are already doing this research, with India becoming the country of most research on this topic, five classes of geology condition occurrence with hard rock class becoming the main focus of studies, and 11 climate conditions with arid and semi-arid becoming the main focus of studies. There is variability of criteria of up to 33 with 18 criteria still used only in single studies and eight dominant criteria have been used; AHP and MIF have become the method most used for assigned weight and FR is the newest method. Lithology has the highest weight compared with the other seven dominant criteria, and well data has become the most-used data for verification of groundwater recharge. As for further research, this finding can be used as reference for finding new novelty in MCDA and GIS for determining groundwater recharge potential.

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

The authors declare there is no conflict.