Spatial analysis of dynamic groundwater potential maps of Nagpur district of Maharashtra using the AHP and RS-GIS tools Open Access

Water availability for the sustainability and economic viability of agriculture-based economies such as India Freshwater, comprising surface, groundwater, and glacial reserves is scarce with groundwater constituting 30% and surface water only 1% of total freshwater (Das & Pal 2019). Groundwater being the second most available freshwater reserve serves most human and economic activities (Doke et al. 2021). The growing population pressurizes groundwater reserves, exacerbated by unscientific planning and over-exploitation, leading to their depletion in India. The Maharashtrian economy relies on agriculture, thus raising critical surface and groundwater issues due to high demands (Pande et al. 2020). Groundwater occurrence is dynamic and naturally replenishable (Sachdeva & Kumar 2021; Upadhyay et al. 2023). Generally, extraction rates surpass replenishment, driven by human intervention, growing population (Tolche 2021), urbanization, agricultural–industrial growth, and unsustainable development (Sachdeva & Kumar 2021), which causes a lowering of groundwater levels (Luong 2021). Thus, timely actions can be taken by installing artificial recharge mechanisms to increase recharge rates.

Major geological formation in Maharashtra state is basaltic rock type that covers approximately 80% of the area (Kumar et al. 2020; Doke et al. 2021). The presence of Deccan traps basaltic rock (DTBR), particularly vesicular basalt and sedimentary formation, encourages groundwater development due to their high permeability (Rai et al. 2013), however, it has variation over the area. Weathered and fractured basaltic formation allows aquifer formation and groundwater storage by secondary porosity. However, water transmission is limited in hard rock basaltic formations due to the absence of inherent pores (Jain et al. 2023). Moreover, basaltic terrain causes post-monsoon water scarcity (Pancholi et al. 2022) due to the presence of deep channels that drain groundwater in unconfined aquifer systems (Yesupogu et al. 2012). Furthermore, rainfall-induced recharge (groundwater reserves, GWR) is a major groundwater source but, the study area experiences semi-critical annual rates, and Katol and Saoner talukas of Nagpur district are facing critical extraction situations (CGWB report 2022). Nagpur district geologically, lies within DTBR formation, comprises multi-layered solidified lava erupted which represents a multi-layered aquifer formation such as sedimentary formations (Rai et al. 2013). Thus, groundwater occurrence in the study area finds favourable conditions but distribution is uneven because of varying rainfall and local water crisis. Limited studies address this problem, several works have been conducted at a micro-scale concentrating on specific locations such as Kalmeshwar and Kamtee. Groundwater potential (GWP) mapping may provide recharge urge in water-scarce areas based on groundwater level fluctuation (GWLF), land use and land cover (LULC), fractional impervious surface (FIS), and rainfall patterns in the past. Furthermore, it may help the installation of suitable techniques for the recuperation of groundwater levels in the future.

The demarcation of GWP can be explored using various methods. Garamhegyi et al. (2020) employed a regression method for multivariate time series analysis for accurate groundwater depletion and recharge estimation in Danube-Tisza Interfluve, Hungary. Hence, in regions where long-term groundwater level data are readily available, statistical methods serve well. Other approaches are deterministic and expert evaluation methods (Saranya & Saravanan 2020) which require spatial representation of point data using interpolation techniques such as kriging, polynomial, inverse distance weighted (IDW), and others techniques. Effective utilization of these three approaches is highly dependent on many factors including the availability of data, desired accuracy, extent of area, and selection of parameters. Movement and storage of groundwater depend on soil type, slope-gradient, topography, lithology, geology, geomorphology, and other groundwater-influencing factors (Greenbaum 1992). Selective consideration of many factors better serves GWP mapping and analysis. However, parameter influence is site-specific, one parameter that contributes more to groundwater occurrence may show a low effect for other locations. Conventional groundwater analysis approaches such as hydrogeological surveys, electrical resistivity methods, electrokinetic system sounding techniques, and radon techniques are time-consuming and very expensive however, they are the most effective means of groundwater exploration. On the other hand, the GWP indexing approach is the most reliable, easy, and commonly used method nowadays. Mapping of GWP zones and their dynamic study can be used to decide the location of dug wells for domestic as well as irrigation supplies. The spatial occurrence of groundwater reserves is highly unpredictable and explicitly relies on spatial variation of geological and geomorphological formations, rainfall–runoff rate, soil infiltration, etc. (Doke et al. 2021), hence its direct measurement is not possible. Thus, spatial analysis of groundwater occurrence becomes explicitly important which facilitates exploration of major groundwater-rich zones and their locations. Current advancements in satellite-based technology allow researchers to demarcate GWP more easily and effectively. Remote sensing and geographical information systems (RS–GIS) emerged as a powerful set of tools used for GWP mapping. The RS–GIS combinedly solves the problem of spatial data acquisition, storage, and management efficiently at a low cost (Tolche 2021). However, their ground truthing is required while accepting the outcomes from RS–GIS results. Remote sensing-based data are unable to detect the presence of groundwater directly below the surface, hence to identify its occurrence, different products of remote sensing such as geomorphology, geology, soil texture, fold and faults, LULC, topographical features, etc. are used as indicators of its occurrence (Todd 1980).

Many GWR influencing factors are spatially combined depending upon their relative importance. Experts' knowledge base is utilized directly or indirectly to decide relative weights. Various studies used analytical hierarchy process (AHP) in combination with RS_GIS for GWP mapping (Roy et al. 2020; Chatterjee & Dutta 2022; Raj et al. 2022; Etuk et al. 2023; Kassa et al. 2023; Silwal et al. 2023). Methods such as fuzzy AHP (Das & Pal 2019; Shekar & Mathew 2023), fuzzy logic, and their combination are also used. Other available methods are influencing factors, frequency ratio, multi-influencing factor (Karimi et al. 2022), random forest, and gradient-boosted trees (Sachdeva & Kumar 2021) for weight assignment. Similarly, many different multi-criteria decision-making methods have been used in the past. The AHP is a simple and effective tool to determine the relative weights of various factors for GWP mapping thus adopted in the present study.

In recent times, different studies have been available working on different sets of parameter layers for GWP mapping. Most common parameters used in these studies are geological (such as geomorphology, lithology, and lineament density (LD)), hydrological (such as drainage density (DD) and rainfall), topographical (as slope, elevation, and digital elevation model), and other variables such as LULC and soil texture (Kumar et al. 2020; Lentswem & Molwalwfthe 2020; Kpiebaya et al. 2022; Pandey et al. 2022; Rajasekhar et al. 2022; Saha et al. 2022). Other studies used digital elevation-derived indices to refine the GWP mapping such as the topographical wetness index (TWI) which represents the availability of soil moisture as a function of topography. Panahi et al. (2020) used many indices such as convergence index, topographic position index, topographical ruggedness index (TRI), and mass balance index along with geology and other catchment parameters to closely identify GWP. The use of proxy parameters for GWP characterization has been extensively used in recent periods. Apart from these, some studies used vertical electrical sounding data for creating layers such as aquifer resistivity, thickness, depth, and vadose zone depth (Rai et al. 2013). However, different combinations of data were used depending on data availability and required precision of work. TWI and TRI along with hydrological, climatic, and topographical data are very less frequently utilized in previous studies. Apart from this, GWLF data that, in a real sense, provides groundwater level variation during the monsoon period were given less importance. Another important parameter known as FIS was less explored for GWP mapping. This parameter represents the extent of imperviousness in a region that may prevent the surface water column from penetrating the ground. Thus, the amount and extent of FIS prevent groundwater occurrence and may cause spatial shifts for specific regions.

The main intention of conducting the present study is to explore the dynamics of GWP mapping in the Nagpur district of Maharashtra. A GWP-based analysis is used in various studies to know the current status of groundwater occurrence in water-scarce areas but an exploration of the dynamic behaviour of GWP is rarely accessed. The present analysis may give useful information in areas where groundwater is reported to be in safe zones such as Nagpur district (CGWB report 2021, 2022). The spatial distribution of groundwater is not exactly predicted with a limited number of observation wells and piezometric heads established by Central Ground Water Board (CGWB). This study will help in understanding the shift of GWP while exploring four main groundwater occurrence factors namely LULC, rainfall, GWLF, and FIS by considering their annual temporal and spatial variation. This study provides a proper knowledge base for decision-makers to identify how and what potential changes have been occurring in the present study area, and it will develop an understanding of future shifts in groundwater depth which will help while drilling new borewells and identify suitable recharging sites. The present study will provide an integrated approach for GWP mapping using the RS–GIS and the AHP exploring spatiotemporal changes in LULC, rainfall, GWLF, and FIS. Parameters are selected based on a rigorous literature survey of past studies. A combination of selected parameters is rarely reported in the literature. This is an attempt to combine the effects of selected parameters including LULC, rainfall, GWLF, FIS, geomorphology, LD, DD, slope, TWI, TRI, and soil type for GWP mapping.

Nagpur is located in northeast Maharashtra, India, historically known as the winter capital extended 20.9369 to 21.7019 latitude to 78.2637 to 79.6507 longitude. 68.31% of its total population (46.53 lakh) resides in urban areas, but urban areas occupy only 4.88% of 9,892 km². It falls within the Vidarbha region comprised of 14 Talukas and 1,859 villages with 13% uninhabited villages (Census 2011). The Vidarbha region is facing a shortage of water supply both for domestic as well as agricultural uses (Rai et al. 2013).

The semi-arid study area receives over 1,000 mm of mean annual rainfall, varying erratically from normal to excess (CGWB 2021, 2022). The majority of monsoonal rain (June–August) is driven by south-western monsoonal winds. The daily maximum temperature in summer is recorded to be 47.8 and 46.2°C in the years 2019 and 2022, but the summer season is found to be more than 45°C temperature in general.

Details of the data used are described in Table 1. LULC maps used have 79.72–84.34% overall accuracy (OA) with overall kappa statistics ranging between 0.421 and 0.589 (Karra et al. 2021). Hence, these maps were found to be suitable for this study.

Fractional vegetation cover (FVC) represents a fraction of vegetation present in the study area. FVC is estimated based on the normalized differential vegetation index. Detailed methods can be referred from Fatema et al. (2023). Maps were reclassified into five groundwater-influencing classes (Table 2).

Here, _NorEle, _MinEle, and _MaxEle represent the mean, minimum, and maximum elevation of the study area, respectively, the TRI map was reclassified into five classes (Table 4) representing their GWP influence.

Pre- and post-monsoon GWLs from 2008 to 2022 were used to derive annual GWLF maps using the IDW interpolation in ArcGIS. The number of measuring stations (MSs) including observation wells and piezometric heads varied, e.g., 69 in 2013 and 95 in 2022. A smaller number of MS was available before 2013. Thus, to address the effect of missing data, the decadal average groundwater level fluctuation (GWLF) was computed while it related to the mean sea level. Consequently, GWLF maps were combined, to derive decadal averaged GWLF maps, then georeferenced and resampled to 10 m resolution. Finally, AGWLF maps were reclassified into five categories (Table 2).

The soil type is a dominating factor influencing GWR in any agriculture-based area (Elvis et al. 2022). ArcGIS 10.5 was used to digitize and georeferenced soil maps obtained from National Bureau of Soil Survey (NBSS) Nagpur. Soil texture (such as clay, fine clay, loam, fine loam, coarse loam, sandy loam, and clayey loam) affects the percolation (Sachdeva & Kumar 2021) and aquifer infiltration. Fine clay and sandy loam occupy over 25 and 40% area, respectively. Different soil types serve differently to GWP. Geomorphology is an important factor that shares a strong link between hydro-geomorphological units and GWR, thus comprehensive inclusion of these helps in the identification of GWP (Upadhyay et al. 2023). The study area occupies 13 different geomorphological features. Predominantly, pediment and pedi-plain occupy more than 78% signifying very high to high GWP. Moderately dissected plateaus (9.42%) and hills and valleys (4.64%) are less important for GWP.

Eleven thematic layers were used to delineate GWP. Relative importance based on Saaty (1987, 2014) was decided based on the rigorous literature review. The relative importance considered in different studies varies at their scale because of different ways of their acceptance such as experts' opinion and AHP-based, current works selected minimum and maximum levels of relative importance accepted in the literature. The final acceptance of the relative importance and weights assignment was based on the following three steps:

• (i)

  Iterating relative importance using a Python program for more than 400 iterations, then finding their weights using the AHP method.

• (ii)

  Constructed pairwise comparison matrices (PCMs) and normalized matrices for each set of iterations. Finally, accepted that relative importance and weights which affirms consistency ratio (CR) less than 0.1 only.

• (iii)

  Taking an average of these weights iterated, affirming to CR less than 0.1.

Non-parametric Mann–Kendall and Sen's slope tests are employed to check whether different GWP categories are showing statistically significant trends (SST) or not. For detailed information on these tests can be referred from Jiqin et al. (2023). These methods are based on two hypotheses, the null hypothesis (H_o) assumes that the concerned category of GWP does not show significant trends whereas the alternative hypothesis (H_a) beholds significantly rising and declining trends in the concerned GWP category. At a 5% significance level, if the p-value is found to be less than 0.05 then only H_a will be accepted. Acceptance of H_o represents that there are no significant trends existing in the concerned GWP category.

Groundwater occurrence in an area is attributed to many factors including climatic factors such as rainfall as well as geological and geomorphological factors. The current study performed GWP mapping by exploiting the dynamic changes of GWLF, rainfall, FIS, and LULC from 2017 to 2022 for the Nagpur district.

Rainfall significantly affects GWR amount and temporal variation of rainfall control infiltration and runoff (Pandey et al. 2022) thus attributed to GWP. Daily rainfall data from the year 2008 to 2022 were collected and annual rainfall maps were derived by summing daily rainfall bands using ArcGIS. Annual rainfall maps are derived by IDW interpolation.

Imperviousness of surface that accounts intern of FIS generally includes the presence of any anthropogenic material that prevents water from infiltrating through the natural surface. As LULC represents a combination of buildings, roads, grassland, waterbody, and barren land which may be subjected to the intermixing of pixels (Lu et al. 2011) thus incorporating FIS leads to a better representation of impervious surfaces and its influence over GWP delineation. Along with LULC, it clears confusion between pixels that are wrongly classified.

Sentinel-2 LULC products were used to assess LULC dynamics and their impacts on GWP. The amount and extent of LULC coverage in an area alters surface properties, enabling rainwater to percolate through soil strata. Urban areas expanded from 485 km² (2017) to 575 km² (2022) which indicates anthropogenic changes Although these changes are marginal (1%) along with other factors, it could modify GWP by altering agriculture practices, changing drainage patterns, and adding imperviousness. It may result in the modification of natural infiltration rates, consequently changing GWR patterns. Vegetative cover plays an important role and controls water infiltration. Higher coverage of vegetation may lead to higher interception with rainfall and may provide an opportunity for water percolation. However, it may also be subjected to a higher amount of evapotranspiration. The study area showed a decrease in cropland (more than 1%) and forest cover (>4%) with a nearly 4% increase in other vegetation including tresses and pasture land (Figure 8). Forested areas could be replaced by pasture lands, flooded vegetation, and barren land which can be observed by looking at their growth in past years.

The dynamics of GWP maps are analysed by putting more emphasis on four key parameters such as rainfall, FIS, AGWLF, and LULC but, other parameters covered in this study play their role and interact with these four differently. The individual role of these parameters is discussed below.

Geomorphological settings are governing factor that facilitates GWR. The geomorphological setting represents the north and western hilly region and Kanchan and Wena river valley areas. The presence of the pediment-peneplain complex and moderately dissected plateau-type geomorphology covers nearly 88% of the total area, which falls under the high to very high GWP category. Thus, geology and geomorphology set Nagpur district within the high to very high GWP zones.

The soil texture is a dominant factor that controls water infiltration. Soils with maximum water infiltration rates facilitate GWR. Generally, sandy and gravel type of soils exhibit high rates of infiltration. The study area is covered with a variety of soils such as sandy loam (40%), fine clay (27), and loam (14). The thickness of soil varies drastically in different parts and ranges between 4 and 100 cm in some places (according to NBSS and Land Use Planning (LUP)). Loamy soils show medium rates of infiltration. Higher depths with loamy soils support higher GWP.

Dynamic effects of DMR, AGWLF, FIS, and LULC along with geomorphology, soil, slope, TWI, TRI, DD, and LD are used to delineate GWP (Figure 18) from 2017 to 2022 using ArcGIS 10.5. Final WOA outputs were categorized into poor, safe, and high GWP. Poor GWP is linked to low TWI, LD, AGWLF, and high FIS, TRI, slope, and DD, alongside finer soils. Coarser soils and other factors facilitate high GWP installation (Tables 2 and 4). In other words, the high GWP category represents the areas capable of storing water as groundwater storage and indicates available groundwater. The high GWP areas are only constrained by topography and hydrological factors. However, they may show minor changes due to other factors (such as LULC) that do not affect the overall productivity of inbuilt aquifer systems. The areas representing gentle slope, nearly plane and mountainous consequent with slow runoff rates, and longer rainwater percolation tendency (Doke et al. 2021) governed high GWP. Safe GWP areas are fragile and bounded by topographical, hydrological, and geological modification along with other direct implications within their extent such as LULC changes. Their installation and degradation cause significant adjustments to aquifer productivity.

The safe GWP category shows continuous setbacks during the study period by losing a significant amount of area to other GWP categories. Statistical trend analysis of various GWP categories, such as Mann–Kendall and Sen's slope test, was performed to explore whether each category shows (SST or not). The results of these tests are presented in Table 5. It was found that the safe GWP category shows SST at a 95% confidence interval, whereas, poor and high GWP categories were not found to have SST at a 95% confidence interval. This may be due to the lower span of data used for this analysis. However, the extent of area under poor and high GWP categories in years 2017 and 2022 represents that both categories are changing. Nagpur city itself falls under the poor GWP category (Figure 15). According to the CGWB report (2021), western talukas of Nagpur district, namely Kalmeshwar and Katol, experience very deep groundwater table patches more than 40 m below ground levels during pre-monsoon season. Some parts of Narkhed, Kalmeshwar, Katol, Savoner maud, Nagpur rural, and Hingna fall within the poor GWP category with a 10–20 m groundwater table below ground level.

Annual rainfall data from 2008 to 2022 indicates mean rainfall exceeding 898 mm yearly. The decadal mean surpasses 1,078 mm, indicating high recharge probability, leading to a concentration of high GWP areas. Masroor et al. (2023) found similar patterns in Parbhani district, Maharashtra, which has nearly the same climatic and geological conditions. The analysis captures the installation of recharge sites while high GWP areas increased from 2017 to 2022. However, the expansion of the poor GWP category was observed in the south-western parts of the district. Kumar et al. (2020) reported similar results while investigating GWP in the Deccan Volcanic Province watershed (in Katol and Kalmeshwar) with 33% experiencing poor GWP. The present study also identified these two talukas under the poor GWP category. Some northern patches indicate increasing high GWP density, showing positive trends, but it poses urgent action for managing GWP.

Along with that, the high GWP area also shows expansion in the southeastern part of the study area (towards Kuhi and Biwapur).

Northeast and south-east parts receive over 1,100 mm decadal rainfall which can be managed more effectively to recharge groundwater. This part is naturally suited for GWP because of low TRI and slope due to relatively flat topography. However, the north and northeast parts are characterized by Satpuda mountains and drained by the Wainganga River in the north or northeast. Moreover, Chandrabhaga and Nag rivers are also flowing from west to east. Hence, the topographical set-up and perennial river flow in the northeastern part of the study area facilitate high GWP. Whereas, the western and south-western portions indicated poor GWP area expansion also covered by Satpuda ranges. The western part and some patches in the southwest show high TRI and slope. This portion receives uneven decadal rainfall ranging from 713 to 1,100 mm. GWLF analysis of this portion shows critical groundwater setbacks. In the year 2017, some wells in this region were receiving immense groundwater recharge after post-monsoon but its intensity was found to be decreased in 2022. So, localized topography, rainfall, and other factors restrict rainfall recharge in this portion. This part of Nagpur district needs immediate GWP management planning and installation of GWR structures.

The present work incorporated three indices, namely TWI, TRI, and FIS, which enabled to capture of minor shifts in the GWP of the study area with other climatic, topographical, and hydrological factors. Especially, the study utilized the dynamics of FIS to identify the real state of imperviousness and its changes in the study area. As the study detected SST in safe groundwater, the study indicated crucial changes occurring recently. The present study shows a requirement for immediate action to manage safe GWP areas in the Nagpur district. All the care has been taken during the study while incorporating the effect of major GWP influencing parameters. Further works may include extensive field data related to soil moisture, terrain, and soil type so that more refinement can be done.

Parameter sensitivity is crucial to identifying parameters that play a measure role for GWP. Eleven parameters were used to develop GWP mapping. To know which parameter is causing maximum changes to GWP, an area-sensitivity approach is used. For this purpose, one parameter's weight is maximized while placing other parameter's weights randomly within the set of weights obtained using the AHP. This step is repeated for all the 11 parameters and their combination. The sensitivity of one parameter is accessed by obtaining the area changes concerning the first GWP map derived using weights based on the literature and the AHP. Several random weights were assigned to other parameters while keeping the maximum weight to the concerned parameter. Finally, the average of that set is taken into account to address the final sensitivity of the concerned parameter. The most sensitive parameters were found to be TRI, slope, soil, geomorphology, rainfall, and TWI causing 35, 29, 20, 19, 11, and 11% sensitivity, respectively, while maximizing their weights. FIS, LULC, and LD were found to be moderately sensitive parameters having 7 to 10% areal changes. GWLF and DD were the least sensitive.

GWP maps were derived using WOA techniques and verified with available GWL data from the India Water Resources Information System (WRIS) portal and CGWB regional office in Nagpur. CGWB measures pre- and post-monsoon GWL in April–June and July–September. Results were validated using pre- and post-monsoon GWL data for May and August incorporating only those wells having both pre- and post-monsoon GWL. Total MS used for validation are 63(2017), 69(2018), 66(2019 & 2020), 63(2021), and 66 (2022), indicating corresponding years in brackets.

The validation process involves creating a vector layer of MS, georeferencing, and intersecting with pre- and post-monsoon GWL and GWP maps. Finally, MS locations were accessed concerning their probable GWL in the GWP category. OA was estimated by taking the ratio of correctly classified pixels to total pixels with wells in a GWP category. OA for various GWP maps is listed in Table 6. It was assumed that GWL should rise after monsoon when results were validated. Additionally, if post-monsoon GWL change is ≤5 m, it agrees with the poor GWP category.

Validating safe and high GWP categories is a challenge due to varying recuperation. Many MS within these categories showed low GWL recovery after post-monsoon, showing barely some rise (<2 m). Only MS with more than 5 m post-monsoon GWL rise were considered. GWL rise between 5 and 10 m indicates safe GWP; over 10 m suggests high GWP. OA was above 64% for all GWP categories, indicating acceptable results. The validation for 2021 was not possible due to a lack of pre-monsoon data likely due to the COVID-19 lockdown. In 2022, no MS were located under the high GWP category, hence no validation was done.

The current study utilized dynamic maps of rainfall, GWLF, FIS, and LULC along with geomorphology, slope, TWI, TRI, DD, LD, and soil maps to explore the trend of GWP of the study area from 2017 to 2022 using the RS–GIS and the AHP technique. The study concludes that there is an immediate need for action to maintain the safe GWP category. The major portion of the study area falls in the safe GWP category and it is showing statistically significant decreasing trends at a 95% confidence interval. This situation is alarming and demands a quick response to prevent further degradation. Moreover, the study area faced around 3,500 km² increase under the high to very high FIS category which restricts the percolation of surface water into the ground. Thus, increasing impervious surfaces in the study area is crucial in modifying GWP occurrence. Furthermore, it will be responsible for generating floods and high runoff in upcoming years. It was also found that during the study period, a significant amount of area having high AGWLF (which means rising GWL) was reduced. It means that during these six years, approximately 10% of the total area lost its high GWR capacity. This situation becomes more critical when TRI and slope are disturbed for any agricultural, industrial, or commercial development, which significantly alters the runoff generation rates in the study area causing urban flooding. Though, dynamics of LULC, FIS, AGWLF, and rainfall are effectively incorporated in the present study to identify potential changes in the GWP of the study area, a micro-scale study with a similar approach may serve better to identify finer shifts in GWP.

The authors express their sincere thanks to USGS by NASA, CGWB, Regional Office, Nagpur, ESRI, IMD, NBSS & LUP, Regional Office, Nagpur and GSI. We are thankful to the Editor in Chief with the team of Water Supply journal and anonymous reviewers for suggesting modifications.

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

The authors declare there is no conflict.