Next Article in Journal
Hydro-Geochemistry and Groundwater Quality Assessment of Ouargla Basin, South of Algeria
Next Article in Special Issue
Controls on Groundwater Fluoride Contamination in Eastern Parts of India: Insights from Unsaturated Zone Fluoride Profiles and AI-Based Modeling
Previous Article in Journal
Research Progress on Integrated Treatment Technologies of Rural Domestic Sewage: A Review
Previous Article in Special Issue
Neural Network and Random Forest-Based Analyses of the Performance of Community Drinking Water Arsenic Treatment Plants
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Prediction of Groundwater Arsenic Hazard Employing Geostatistical Modelling for the Ganga Basin, India

Department of Hydrology, Indian Institute of Technology Roorkee, Roorkee 247667, Uttarakhand, India
*
Author to whom correspondence should be addressed.
Water 2022, 14(15), 2440; https://doi.org/10.3390/w14152440
Submission received: 15 June 2022 / Revised: 3 August 2022 / Accepted: 3 August 2022 / Published: 6 August 2022

Abstract

:
Elevated arsenic concentrations in groundwater in the Ganga–Brahmaputra–Meghna (GBM) river basin of India has created an alarming situation. Considering that India is one of the largest consumers of groundwater for a variety of uses such as drinking, irrigation, and industry, it is imperative to determine arsenic occurrence and hazard for sustainable groundwater management. The current study focused on the evaluation of arsenic occurrence and groundwater arsenic hazard for the Ganga basin employing Analytical Hierarchy Process (AHP) and Frequency Ratio (FR) models. Furthermore, arsenic hazard maps were prepared using a Kriging interpolation method and with overlay analysis in the GIS platform based on the available secondary datasets. Both models generated satisfactory results with minimum differences. The highest hazard likelihood has been displayed around and along the Ganges River. Most of the Uttar Pradesh and Bihar; and parts of Rajasthan, Chhattisgarh, Jharkhand, Madhya Pradesh, and eastern and western regions of West Bengal show a high arsenic hazard. More discrete results were rendered by the AHP model. Validation of arsenic hazard maps was performed through evaluating the Area Under Receiver Operating Characteristics metric (AUROC), where AUC values for both models ranged from 0.7 to 0.8. Furthermore, the final output was also validated against the primary arsenic data generated through field sampling for the districts of two states, viz Bihar (2019) and Uttar Pradesh (2021). Both models showed good accuracy in the spatial prediction of arsenic hazard.

1. Introduction

Fresh water is essential for survival. It is generally understood that surface water would be more prone to contamination due to industrial, municipal, and agricultural discharge necessitating its treatment before use, whereas groundwater is less affected by pathogens. Therefore, dependency on groundwater has increased, resulting in its over-exploitation and water quality deterioration due to various anthropogenic and geogenic factors. Arsenic toxicity has been acknowledged as a significant global groundwater concern, adversely affecting humans even at low concentrations. CGWB [1] has reported 10 states of India as arsenic hotspots. It is observed that a significant portion of the flood plains of the Ganges and Brahmaputra basins have been reportedly contaminated by arsenic. Several hypotheses have suggested the importance of the geogenic release of arsenic from weathered igneous and sedimentary rocks under oxidizing and reducing conditions, respectively [2]. Naturally, arsenic is a significant constituent of more than 100 minerals such as ore including native arsenic, arsenopyrite, arsenic-bearing sulfides, erythrite, and enargite [3]. Therefore, the parent rocks mark the primary source of arsenic in groundwater through different processes such as weathering followed by subsequent leaching and runoff [4,5,6]. Anthropogenic activities related to growth in urban, agricultural, and industrial sectors (e.g., mining, smelting, refining, and other industrial processes) may also contribute arsenic to groundwater [6,7,8]. Similarly, microorganisms also transfer arsenic through bioaccumulation and biotransformation, serving as additional biogenic drivers [9,10,11,12].
Arsenic is a metalloid that exists in both organic and inorganic forms. Inorganic arsenic (iAs) exists in the oxidation states of arsines (As3−), elemental arsenic (As0), arsenite (As3+), and arsenate (As5+), which are widely distributed in the Earth’s crust in association with iron and sulphides. Out of these iAs forms, arsenite (As3+) and arsenate (As5+) are highly toxic and mobile [13,14]. Microorganisms play a crucial role in the oxidation, reduction, methylation, and demethylation of arsenic species affecting their mobility and speciation in the environment [15,16,17]. In oxygenated water, arsenic is present mainly as arsenate (+5), while under reducing conditions it is present as arsenite (+3) [18]. In contrast, microbially converted organic forms of arsenic are methylarsonic acid (MMA), dimethylarsinic acid/cacodylic acid (DMA), arsenobetaine, and arsenocholine, which in terms of toxicity, are considered non-toxic [19].
Drinking water is one of the main pathways of arsenic exposure in people, and the acceptable limit of arsenic in drinking water in India is 0.01 mg/L [20]. There are a few reasons behind selecting the study area comprising arsenic-affected districts of the Ganga Basin. First, according to CGWB reports, the dominant arsenic-affected states marked as arsenic hotspots lie in the vicinity of the main stem of the Ganga River. Second, a significant portion of the Ganga basin is occupied by highly productive agriculture, usually irrigated with arsenic-contaminated groundwater, which has reportedly reduced the production of cash crops such as sugarcane, cotton, and oilseeds. Finally, the population residing around and in the vicinity of the Ganga river has also demonstrated the adverse impact of arsenic on their health [1,15].
Earlier studies reported by various researchers used machine learning to employ the random forest model, regression tree model, logistic regression model, etc., for modeling arsenic in groundwater [21,22,23,24,25,26,27,28]. Podgorski and Berg evaluated the global arsenic threat, Mukherjee et al. and Podgorski et al. worked on pan-India arsenic contamination, and the others worked on the states of India, such as Assam, Bihar, Gujarat, Uttar Pradesh, etc. [21,22,23,24,25,26,27,28]. Considering an apparent gap in the area of mathematical and geostatistical modeling for arsenic contamination, the present study has been undertaken to study the occurrence of arsenic in groundwater and categorize the regions of the Ganga Basin in terms of the degree of hazard likelihood employing geostatistical modelling on the basis of available and collected hydrogeological and hydro-chemical data.

2. Materials and Methods

2.1. Study Area

The Ganga Basin is the largest river basin in India, supporting over 500 million people, out of which tens of millions of people are exposed to arsenic from groundwater [29]. The Ganga River basin (GRB) is a part of the Ganga–Brahmaputra–Meghna River basin, covering 1.086 million square kilometers (km2) of total land area with an estimated mean annual river flow volume of over 550 billion cubic meters (Bm3). The GRB area covers 14% of India, with the late quaternary deposits bestowing their riches in and around the Ganges [30]. The basin extends from the semi-arid Himalayan valleys in the north to the densely forested mountains, Shiwalik foothills, and fertile Gangetic Plains in the south [31]. The fluvial plains are spread throughout the vicinity of the Ganga, Yamuna, Chambal, Son, Betwa, and Ken rivers, which stretch up to 2510 km, 1376 km, 1024 km, 784 km, and 427 km, respectively [32]. The southern bank of the Ganga River comprises flood plains, which are characterized as low-lying mudflats. The GRB is densely populated and is home to 8.3% of the world’s 7.6 billion population [30]. Due to the richness of sand, loam, clay, and their combinations in the entire basin, most of the population depends upon agriculture for food and livelihood, which entirely depends on water availability from the rain and the Ganges [31,33] (Figure 1).

2.2. Methodology

The first step was to identify the variables or factors (such as subsurface geomorphology, groundwater table depth, water quality, etc.), which plausibly have a direct or indirect association (to be established statistically) with arsenic mobilization and can be quantified. The second step envisaged creating a platform having aggregated numerical values corresponding to the identified variables employing a pre-decided computational framework yielding the hazard potential. The third step was the overall presentation on a graphical (GIS) platform for the affected Ganga basin states [35]. The available secondary data was collected from various sources and analyzed mathematically and statistically. Analytical Hierarchy Process (AHP) and Frequency Ratio (FR) methods were employed to calculate and assign weights, ratings, and prediction values to each covariate [35,36,37,38]. Finally, the final groundwater arsenic hazard map was generated using a raster calculator and weighted overlay analysis in GIS.
The schematic diagram of the methodology applied is presented (Figure 2) and is divided into the following main themes:
  • Data procurement and preparation (including GIS-based database);
  • Covariate analysis and Geostatistical Modeling for developing the Arsenic Hazard Index (AHI);
  • Generation of groundwater arsenic hazard maps and validation.

2.2.1. Data Procurement and Preparation

Initially, the selection of variables was performed based on existing information and literature related to arsenic contamination in groundwater. Later, secondary data for the selected variables were procured from the official documents, websites, and reports prepared by various academic/research institutes and government and non-government organizations (Table 1). All the procured datasets were further processed in GIS, a simple platform for complicated and dense datasets. Currently, GIS has a wide application in problem-solving and decision-making processes [39]. The final arsenic hazard map was generated using an overlay analysis tool in ArcGIS v. 10.6.1. All the datasets in 2D and 3D surfaces were represented with the suitable map projections called the Universal Transverse Mercator (UTM).
The depth of water levels were procured for the pre-monsoon season, and the map was created using the Ordinary Kriging Geostatistical tool of ArcGIS. A total of 10 Shuttle Radar Topography Mission (SRTM) images of 30 m resolution available in TIF format were processed using the spatial analyst tool to create the slope map. Furthermore, all the split images were merged to produce a single DEM of the study area using the Mosaic tool in ArcGIS, followed by the image enhancement. For the geological database, such as geomorphology and types of aquifers, the maps were geo-referenced by adding four corner coordinates in ArcGIS software, followed by digitization. The acquired global TIF layer was clipped to generate a soil map for the required area, which was further projected into UTM.
Similarly, 10 Landsat 8 images were merged and clipped, followed by supervised classification to distinguish the LULC classes. The rainfall data were also procured for the pre-monsoon season and used after reclassification. After procuring the groundwater abstraction database, kriging was performed to generate the thematic layer [46]. Based on their statistically established correlation with arsenic, a few groundwater quality parameters, such as dissolved silica, bicarbonates, iron, conductivity, hardness, and sulphates, were selected. Groundwater arsenic concentration datasets for the current study were divided into training (75%) and testing datasets (25%). The training datasets were used to run the model, while the testing datasets were used for assessment of the arsenic occurrence and validation.

2.2.2. Variable Analysis and Geostatistical Modelling for Developing the Arsenic Hazard Index (AHI)

The following sections describe the details pertaining to the analysis of covariate and arsenic data for establishing the underlying association, the AHP and FR models, and the development of the AHI [35].

Assessment of Association of Variables with Arsenic

After preparing the individual thematic layers for all the selected variables and arsenic, their geostatistical association was evaluated by overlay analysis for each covariate layer with the testing datasets considering the area percentage of each class of covariate (Kriging interpolation) and the percentage of arsenic occurrence (the number of locations in each class) and was represented using line graphs.

Analytical Hierarchy Process (AHP) Model

The AHP is a multi-criteria decision method that uses hierarchical structures to represent a problem and then develops priorities for alternatives based on the user’s judgment [47]. “It facilitates decomposition, and pair-wise comparison reduces inconsistency and calculates priority vectors” [48]. This method involved the following steps:
  • The unstructured problem was defined.
  • The hierarchical framework was developed based on the selected variables.
  • The pair-wise comparison matrix between all variables was constructed.
  • Based on Saaty’s scale (Table 2), weights were assigned to all variables at each hierarchy level using the eigenvalue technique [49].
  • The ranking for pair-wise comparisons was performed based on the reviews from literature to obtain the relative importance of the alternatives.
  • Furthermore, consistency verification was performed using the following equations after estimation of the eigenvalue for each row:
    • The consistency index (CI) was calculated as below,
      CI = ( λ max     n ) ( n 1 )
      where λmax = Maximum Eigenvalue of the matrix
      n = number of groundwater affecting variables.
    • The consistency ratio (CR) was calculated as below:
      CR = CI RCI
      where RCI = Random Consistency Index (Table 3).
  • All the steps from 3 to 6 were repeated for each hierarchy level. It was ensured that the CR values must be less than 0.1 [47].
  • Finally, an overall priority ranking was developed for selecting the best alternative.

Frequency Ratio (FR)

The frequency ratio is a bivariate statistic approach to explain the prospect of the occurrence of a certain covariate [50,51]. It considers both the dependent and independent variables to determine their relationship [52]. In this context, the FR model was used to predict the temporal changes by assuming that the associated variables determine the future arsenic occurrence under the same conditions as in the past [53]. Hence, the FR was calculated by the correlation between the prediction sites and variables using the equation:
FR = A / A B / B
  • where FR = frequency ratio of a class for the covariate,
  • A = the number of pixels with arsenic hazard for each covariate,
  • A′= the total number of pixels with arsenic hazard,
  • B = the number of pixels in the class area of the covariate, and
  • B′= the number of total pixels in the study area.
Based on the frequency ratio, the relative frequency was estimated by normalization, followed by assigning the prediction rates to each covariate [54]. PR was calculated as:
PR = ( RF max RF min ) ( RF max RF min ) min
  • where PR = the prediction rate,
  • RF = the relative frequency,
  • Max = the maximum value of RF, and
  • Min = minimum value of the RF.
FR values of more than one show a strong correlation with arsenic and were considered the most arsenic contaminated and vice versa [53,55].

Arsenic Hazard Index (AHI)

The AHI is a dimensionless quantity applied in the current study to map groundwater arsenic hazard. The AHI demonstrates the intrinsic geological and hydrogeological characteristics of a site to determine the hazard likelihood of groundwater arsenic contamination. “The risk of groundwater contamination is defined as the probability that groundwater becomes contaminated to unacceptable levels, using the activity that occurs directly on and below the earth’s surface” [56]. All the individual thematic maps were combined with the groundwater arsenic concentration layer for the current study. The AHI was calculated by the weighted linear combination (WLC) technique using [35]:
AHI = Σ PrPw
  • where P = the individual covariate,
  • r = rating (AHP) and prediction rates (FR), and
  • w= weightage (AHP) and relative frequency (FR).
The AHI values were estimated using a raster calculator in ArcGIS and grouped under four classes—very low, low, moderate, and high—based on the natural jerk classification method of ArcGIS.

2.2.3. Generation of Groundwater Arsenic Hazard Maps and Validation

The final output map was prepared by overlay analysis in ArcGIS 10.6.1. Several attempts were made employing spatial analyst tools, raster calculator, and overall weightage calculation and representation. The final hazard map was the resultant of an overlay of all the thematic layers with specific rates and weights for the AHP. For FR methods, rates were assigned based on ratios calculated.
To determine the accuracy of the resultant, data validation is essential. The validation exhibits how well the models predicted groundwater arsenic hazard in the study area. The current study focused on data validation in two ways:
  • By AUROC: The area under the Receiver Operating Characteristics (AUROC) curve was plotted between the sensitivity (true positive rate, TPR) and specificity (1–FPR). TPR is the fraction of interpretations that were correctly predicted to be positive out of all positive interpretations (TP/(TP + FN)), while the false-positive rate (FPR) is the fraction of interpretations that are incorrectly predicted to be positive out of all negative interpretations (FP/(TN + FP)). It tells us the accuracy of the model applied. This is the independent cut-off metric that has wide application in hydrological studies. AUROC was performed using the ArcSDM tool in GIS, which required a final output map and testing datasets (25% of total arsenic samples). AUROC may vary from 0 to 1; the higher the value, the higher the reliability [57,58].
  • Through Current Scenario: 201 groundwater samples were physically collected from three districts each of Bihar and Uttar Pradesh to validate the predicted arsenic hazard with the ground situation (sampling was performed in Dec 2019 (pre-COVID) and March 2021 (post-COVID), respectively). The selection of states and districts was wholly based on the recent groundwater arsenic hotspot declarations [1]. The standard APHA procedures were followed during the collection of samples and field measurements [59]. Groundwater samples were collected from the handpumps at the locations as mentioned in the secondary arsenic datasets in pre-washed plastic bottles. Before sampling, each hand pump was purged for about 10–15 min. Water temperature, pH, EC, oxidation-reduction potential (ORP), and As concentration were measured at the site using portable analysis kits (HACH HQ40d, MQuant Arsenic Test Kit), whereas other ionic species and heavy metals analysis were performed with the instruments (Metrohm Ion Chromatograph, Agilent 8900 Triple Quadrupole ICP-MS) available in the institute laboratories.

3. Results and Discussion

Rocks, being the key reservoir of arsenic with the significant contribution of arsenic-bearing minerals, mark the geogenic sources as the most accepted theory for the occurrence of arsenic in the groundwater. In the Indian context, it is well stated that the sediments from the Himalayas are the primary sources of arsenic in groundwater of the Ganga–Brahmaputra–Meghna plains [60]. In the current study, elevated arsenic concentrations were noted around the regions of alluvial plains along the Ganges River, confined by the sand, silt, and clay layers [61]. The As-bearing minerals in the solid phase apparently released arsenic into groundwater in these affected regions upon facing favourable redox conditions. The study findings are presented and discussed in the following sections:

3.1. Assessment of Association of Arsenic with Variables

To evaluate the association of arsenic with the variables, the testing datasets were categorized into three classes, i.e., <0.01 mg/L, 0.01–0.05, and >0.05 mg/L [35]. This classification was based on the permissible limits of arsenic in the drinking water in India, which is 0.01 mg/L, but in the absence of an alternative, it can be increased to 0.05 mg/L [19,20]. The observations regarding the justification for selection of identified variables and their geospatial association with arsenic are presented below:

3.1.1. Depth to Water Level (DTW)

The depth to water level map showed that the water levels are deeper in the northern part of Rajasthan (Jaipur, Dausa, and Sikar districts), middle parts of Haryana (Karnal, Panipat), West Bengal (Bardhhaman, Hugli), and south Delhi. In contrast, in the middle Ganga plains, low surface elevation and shallow water level (<10 m) were observed (Figure 3a). Elevated arsenic concentration was observed in the shallow regions of the Ganga basin. Our findings support the observation of the decrease in arsenic contamination with depth [62].
The line graph also shows that shallow depths in the range of 4–7 mbgl have high arsenic due to the release from oxidation of arsenic-rich iron sulphide minerals (Figure 3a) [63]. In contrast, less arsenic in deep aquifers could be due to the hard older rocks with less or no organic carbon (no biological activity). Second, arsenic could be absorbed into the sediments it passed along the way to deep depths. One finding also proved that a thick clay layer in the upper aquifer prevented the downward movement of arsenic [64].

3.1.2. Slope

As per the spatial variation of arsenic with slope (Figure 3b), a significant portion of the basin has a low slope and elevated arsenic. Areas with low slopes (low surface elevation) retain water for long durations, resulting in the accumulation of fine sediments of arsenic adsorbing minerals and increased infiltration, releasing arsenic in groundwater [65]. Hence, the slope and arsenic distribution are inversely related. Likewise, regions with steep slopes experience more runoff with less infiltration and hence, are less prone to groundwater arsenic contamination. It is also observed that slope plays an important role in controlling hydraulic gradients and groundwater flow. Due to their deposition under high flow regimes, coarser grained sands are found at higher elevations, which promote flushing rates due to the development of higher hydraulic heads and limit the inflow of arsenic-enriched groundwater from the fine-grained low-lying areas due to the potentiometric barrier. Hence, low slopes typically lead to low arsenic flushing rates [66]. The findings show that infiltration is apparently one of the sources of arsenic in the groundwater, which has, however, the least contribution (only 2%) among the variables. Still, it cannot be ignored.

3.1.3. Geomorphology

The basin is covered with alluvial plains (younger, older, and flood plains) and pediment pediplain. Since 43.5% of the area is covered by alluvium, arsenic contamination is anticipated in the alluvial and flood plains of the basin (Figure 3c). Arsenic distribution in the basin showed a considerable association with the quaternary sediments of alluvium plains. The mineralogical composition of these sediments is quartz, feldspars, illite, and the fine-grained over bank feces, which is rich in organic matter [67,68,69]. Similarly, major arsenic hotspots were observed in and around the abandoned/cut-off palaeo-channels of the Ganga, deep flood plains with mud/clay deposits enriched with organic carbon. Active flood plains usually have curvilinear depressions (due to surrounded water bodies) and elevated arsenic concentration because of sedimentation, which releases arsenic into the groundwater. Arsenic becomes trapped as a coating on organic matter and sediments, which is later released by the reductive dissolution of iron oxide and subsequent oxidation of organic matter [70]. Hence, geomorphology represents the geogenic source of arsenic in groundwater.

3.1.4. Types of Aquifers

Types of aquifers deal with the hydrogeology of the basin. The study area is dominated by an unconsolidated sedimentary aquifer, extending its stretch along the Ganges River (Figure 3d). Unconsolidated aquifers with unstratified layers of alluvial sediments originated from the Himalayan catchments. Generally, these sand and gravel sediments have high porosity, hydraulic conductivity, and transmissivity, resulting in the release of arsenic into groundwater by the reductive dissolution mechanism of iron oxides and hydroxides [71].

3.1.5. Soil Types

Arsenic absorption behaviour governs arsenic release in soil and sediments [72]. The basin is primarily covered with alluvial plains composed of sand, silt, and clay (60%) in varying proportions, i.e., older alluvial of Pleistocene (2.6 million to 12 million years before the present) and younger alluvial of Holocene (11,700 years to present) with the sediments deposited from the Ganges River and its tributaries. The alluvial stretch extends from Rajasthan through the middle basin covering Haryana, Delhi, Uttar Pradesh, and Bihar up to West Bengal (Figure 3e). Iron oxides promote arsenic mobilization by forming arsenic-containing pyrites that later undergo oxidation and discharge arsenic by reducing iron oxyhydroxide under anoxic conditions [73,74]. Hence, oxidation of pyrites and reductive dissolution of iron oxides and hydroxides during sediment burial apparently favoured leaching of arsenic from the soil [67,75].

3.1.6. Land Use Land Cover (LULC)

Land use land cover is the description of land-use patterns. The Ganga basin has various land-use patterns, including 74% of croplands (Figure 3f). Hence, most land in the basin is highly cultivated and fertile. Furthermore, forests stretch up to 14% of the basin area. Elevated arsenic was observed in the agricultural area due to the accumulation and absorption of arsenic in soil and sediments. Irrigation with arsenic-contaminated water understandably added to it, which found its way through the food chain. Croplands have high organic matter (OM) content, enhancing arsenic release from soil sediments into the soil solution, further promoting arsenic leaching into the groundwater. Anawar et al. [76] revealed that OM acts as the main redox driver in the reductive dissolution of iron hydroxides which accelerates the mobility of arsenic and hence its release. It was also observed that OM strongly influences microbial-mediated redox reactions, adsorption, desorption, and complexation reactions.

3.1.7. Rainfall

Rainfall is the principal source of recharge of groundwater. For the current study, rainfall for the pre-monsoon season was considered based on the availability of water quality data (Figure 3g). Low or negligible rainfall in most parts of the basin was observed during this season. Low rainfall accompanies high arsenic accumulation due to less mobility, adsorption, and low flow rate [77]. Conversely, a high recharge rate, either naturally or by artificial recharge, reduces arsenic contamination in groundwater. During the monsoon, it was observed that a rise in the hydraulic head promotes the groundwater flow and dilution of arsenic in groundwater. Heavy rain leads to flushing of aquifers through the fissures or fractures in arsenic-bearing rocks by infiltrating rainwater into the aquifer, resulting in arsenic dilution [78,79]. Groundwater recharge also lowers arsenic sorption and release into groundwater by inhibiting the mineralization of dissolved organic matter (DOC). Higher DOC mineralization, on the other hand, may result in increased arsenic sorption, which is then released into groundwater via the reductive dissolution of iron oxide [72,80].

3.1.8. Groundwater Abstraction

Nowadays, a major portion of groundwater is used for drinking and irrigation. It was observed that 63% of the total basin’s groundwater is abstracted on a regular basis (Figure 3h). Heavy pumping for irrigation from an aquifer reduces the compactness of adjoining clay and water, resulting in land sinks, which may contain arsenic, forcing itinto deep aquifers. Hence, high abstraction accompanies a lowered water table and the oxidation of pyrite, liberating arsenic in deep aquifers [68].

3.1.9. Groundwater Quality

Arsenic exhibited both positive and negative correlations with groundwater quality parameters (Table 4). It was observed to be positively correlated with silica, bicarbonates, and iron, whereas a negative correlation was displayed with electrical conductivity, hardness, and sulphate.
The association of arsenic with quality parameters is described as follows:
  • Dissolved Silica
    Weathering of hard rock aquifers results in the release of silicates in groundwater, which is available in the form of dissolved silica (SiO2) [81]. As per the data, silica concentration ranged from 1 to 170 mg/L (Figure 3i). Higher silica was reported in the central basin states such as Uttar Pradesh, Madhya Pradesh, and northern Jharkhand. However, the highest concentration of arsenic was reported in Bihar, Uttar Pradesh, and West Bengal (>0.05 mg/L). These districts also lie in alluvial plains, where arsenic becomes trapped and creates reducing conditions, followed by the reductive dissolution mechanism to release arsenic into groundwater [70]. Silicate minerals from the bulk of sediments marked the largest arsenic reservoir with 75% arsenic, whereas Fe–Mn oxyhydroxides, the minor elements, made the second-largest arsenic reservoir with 16% arsenic [82]. This study observed a high percentage of arsenic occurrence in the regions of dissolved silica from 23 to 34 mg/L, suggesting that high silica promote high pH and low redox conditions accompanying silicate mineral dissolution, which in turn releases arsenic from silicate minerals.
  • Bicarbonates
    Groundwater bicarbonate concentration ranged from 61 to 860 mg/L. Higher bicarbonate concentration was reported in the states of Bihar, Rajasthan, Uttar Pradesh, and West Bengal, whereas a similar pattern was observed with arsenic (>0.05 mg/L) (Figure 3j). The shallow alluvial stratigraphy enriched with organic carbon permits a part of it to infiltrate into deep aquifers with the percolating water. This organic carbon promotes microbial respiration and reductive dissolution resulting in the leaching of arsenic and HCO3 into groundwater [61]. This study observed a high percentage of arsenic occurrence in the regions with bicarbonate concentrations ranging from 323 to 499 mg/L, suggesting that bicarbonates have the ability of complex formation with the iron and manganese oxyhydroxide, which results in substituting bicarbonates for arsenic in the minerals and sediments and releases the arsenic into the groundwater.
  • Iron
    Groundwater iron concentration ranges from 0.05 to 12.51 mg/L. Higher iron concentration was reported in southern Haryana, West Bengal; northern Chhattisgarh, Jharkhand; northeastern parts of Madhya Pradesh; and distributed uniformly in Bihar. The majority of the area with iron concentration 0.88–2.88 mg/L had a high concentration of arsenic (>0.05 mg/L) (Figure 3k). During the dry season, the chemical weathering of clayey sediments loses Na and K; hence, less mobile elements such as Fe and Al remain enriched. Due to the strong affinity of arsenic with pyrites, the weathering of pyrite releases arsenic into the groundwater. This study showed a high percentage of arsenic occurrence in the regions with iron concentrations ranging from 0.89 to 2.88 mg/L, suggesting that reductive dissolution of Fe–Mn oxyhydroxides release arsenic in groundwater [82].
  • Electrical Conductivity (EC)
    Electrical conductivity is one of the groundwater quality parameters dealing with charged particles and dissolved solids in the groundwater. High conductance attributes of Na+, K+, Mg2+, Ca2+, Cl, SO42−, and HCO3 ions in groundwater have a high binding affinity towards arsenic and hence release less or no arsenic in groundwater [83]. As most of the study area comprises alluvium, high EC in some parts could be because of ion exchange and solubilization with agricultural runoff [84]. In the present study, EC values ranged from 141 to 3060 μS/cm with values predominately ranging from 141 to 851 μS/cm (Figure 3l). The study represents low EC values in the pre-monsoon season with elevated arsenic, suggesting less binding affinity of ions towards arsenic, resulting in the release into the groundwater.
  • Hardness
    As per the data, total hardness concentration ranged from 64 to 2779 mg/L. Elevated hardness concentration was reported in 0.3 % of the area, including Haryana, Rajasthan, Uttar Pradesh, and West Bengal, which had the least arsenic testing data points. Hardness ranging from 256 to 405 mg/L occupied more area (approx. 45%), and most of the data points in these locations also represented a high concentration of arsenic (>0.05 mg/L) (Figure 3m). The results showed that low hardness (Ca2+ and Mg2+) increases dissolved iron concentration, undergoes reductive dissolution, and leaches arsenic into the groundwater.
    Conversely, it is stated that high hardness enhances the binding capacity of Ca and Mg ions to form arsenate complexes, resulting in less mobilization of arsenic [85,86]. A finding stated that sorption of Ca and Mg ions increase positive surface charge, increasing its valency and increasing anion sorption such as arsenate ions onto iron oxides through charge neutralization [87,88]. Hence, in the current study, low hardness is observed accompanied by low adsorption and increased Fe hydroxide flocs, which boost co-precipitation and release of arsenic [89].
  • Sulphate
    Sulphate concentration ranged from 2 to 745 mg/L. Higher sulphate concentration was reported in the states of Haryana, northern Himachal Pradesh, northern parts of Madhya Pradesh, Rajasthan, southern Uttar Pradesh, central Jharkhand, and western West Bengal. Most of the data points at the locations with concentrations of 14 to 29 mg/L displayed a high concentration of arsenic (>0.05 mg/L) (Figure 3n). Low sulphate with high arsenic reflects low redox potential (more contamination), promoting sulphate reduction and inhibiting oxidation. Low sulphate accelerates reactive iron concentration, which in association with microbes results in the reductive dissolution of arsenic-rich iron oxyhydroxide [90,91]. This reaction is accompanied by biogenic pyrite formation, increased arsenic mobility, and its release into groundwater [92,93]. This association supports the negative correlation of arsenic with sulphate.
  • Arsenic
    Training datasets (75%) of arsenic concentration were used to generate the thematic layers and for geostatistical modelling. As the arsenic concentration ranges from 0 to 1 mg/L, the training and testing datasets were distributed in three classes, as per the permissible limits: <0.01, 0.01–0.05, and >0.05 mg/L. Training datasets showed higher arsenic in Bihar, Jharkhand, and West Bengal. Most of the data points at these locations matched the high concentration of arsenic in the testing datasets (>0.05 mg/L) (Figure 3o). The mechanism involved in arsenic release apparently includes the oxidation of pyrite, the reductive dissolution of iron oxyhydroxide (bacteria-mediated reductive dissolution), desorption/absorption, sulphide oxidation, and competitive ion exchange [68,94,95,96].

3.2. Groundwater Arsenic Hazard Mapping

  • Assessment of weights and ratings and evaluation of prediction rates:
    Based on the AHP model, the weights and ratings for all the variables were evaluated. The assessment was wholly based on the correlation coefficient values of arsenic with each covariate and within each covariate class. Computation modalities of weights and the rating for the selected variables are explained in Table S1. Pair-wise comparison specified that water depth, geomorphology, hydrogeology, and land-use patterns had the highest weightage. The overall consistency ratio for the data used was 0.08 (<1), reflecting reasonable consistency for the present study.
    Similarly, FR values were calculated with each thematic layer’s prediction rate, based on the relationship of each layer with arsenic testing datasets. The tabular format for the same is provided in Table S2.
  • Computation of Arsenic Hazard Indexes and generation of maps:
    After assigning weightage and rating to the variables based on the AHP, all the thematic layers were combined into a single layer using the arsenic hazard index and overlay analysis in ArcGIS. The AHI was computed as the sum of the product of weights and ratings assigned to each covariate considered for the study [35]. The generated final arsenic hazard map correctly visualizes the likelihood of arsenic hazard zones. The index values are divided into four categories in the order of increasing degree of hazard likelihood, viz, very low, low, moderate, and high [97,98]. The final map is presented with natural the Jenks classification method in ArcGIS, based on the natural groupings inherent in the data [97] and shown in (Figure 4a). The map represents a discrete pattern. The highest index value has been displayed around and along the Ganges River, which lies in the channel of the alluvium plains. The majority of Uttar Pradesh, Bihar, parts of Rajasthan, Chhattisgarh, Jharkhand, Madhya Pradesh, and eastern and western regions of West Bengal show a high arsenic contamination of approx. 35% (Table 5, Figure 5). These highly contaminated zones require more attention towards better land and water resources management in order not to further raise arsenic pollution and toxicity among the inhabitants.
Groundwater arsenic hazard mapping was also carried out using the FR model, where the frequency ratio and prediction rates were assigned to all the individual variables. Later, all the thematic layers were combined using the evaluated prediction rates and the overlay analysis tool. The output map represented a smooth pattern with the natural jerk classification intervals (Figure 4b) [97]. All the values were observed to be above one, indicating that all the variables showed a positive correlation with arsenic. Among all the variables, the relationship between geomorphic features and groundwater arsenic contamination had the highest FR value of 13.4. The slope showed the least FR value of 1.2, showing a decreasing trend with an increasing slope. Depth showed a significant association with arsenic hazard, and the same trend was observed as for slope. In the hydrogeology domain, unconsolidated sedimentary rocks showed an FR of 1.8, responsible for high arsenic contamination in groundwater due to leaching. Due to the huge population and domestic activities, built-up areas apparently were responsible for more contamination, with an FR value of 3.4. The FR values for groundwater quality parameters lie between 1 to 5. From the FR values, the prediction rates were calculated, where slope held the highest predictability with 4.33. The central part of the basin showed a high arsenic hazard in the alluvial stretch around the Ganges River, with shallow depth, low slope, high groundwater abstraction, high dissolved silica, bicarbonates, and iron but low conductivity and sulphate concentration. High hazard from the FR model comprised 37% of the study area with a smooth and continuous pattern (Table 5, Figure 5). In contrast, the results of the AHP method represented a discrete and maybe an actual scenario. It was also observed that the final arsenic hazard maps obtained are quite similar to the recent study of arsenic prediction through machine learning, except for Uttar Pradesh by Mukherjee et al. and Podgorski et al. [23,24]. Therefore, the results obtained using both methods could be adjudged as satisfactory and successful.

3.3. Validation

Validation is an essential process of modelling to assess the accuracy of the model. The study follows two validation methods:

3.3.1. By AUROC Curve

The Receiver Operating Characteristics metric was used to evaluate the output quality using cross-validation. Steepness and Area under Curve (AUC) are very important for validation, as they are perfect for maximizing and minimizing the true-positive and false-positive rates, respectively. ROC curve analysis was computed using arsenic training/testing datasets and the arsenic hazard map obtained by the AHP and FR models (Figure 6a,b) [99,100].
Based on the prediction accuracy and AUC values, AUROC was divided into the following classes: 0.9–1.0 (excellent), 0.8–0.9 (very good) 0.7–0.8 (good), 0.6–0.7 (average), and 0.5–0.6 (poor) [101,102,103]. AUC values for both training (used for running model) and testing (not used in the model) datasets were used to measure success and prediction rates. AUC values range from 0.7 to 0.8 (close to 1) in the study. Therefore, both models could be stated to show good accuracy in the spatial prediction of groundwater arsenic hazard.
Though the values for both methods are almost the same for AUC, various research studies have reported FR to be more accurate as it considers the frequency of arsenic occurrence [37,52,104]. Simultaneously, the AHP concentrates more on the factor responsible for arsenic hazard in the study area.

3.3.2. Comparison with Actual Field Data

The accuracy of the model was also determined by validating the predicted hazard with groundwater arsenic concentrations of three districts each of two states, Bihar and Uttar Pradesh (Figure 7). A total of 93 water samples from handpumps of three districts of Bihar named Saran, Samastipur, and Vaishali were collected in 2019 (pre-COVID period). While comparing both secondary and primary arsenic concentrations, it was found that during 2015 these three districts had a low hazard. However, predictions from the current study highlighted a higher incidence of arsenic contamination. The same was observed with the primary data collected in 2019.
A further 108 water samples were collected from hand pumps of three districts of UP named Ballia, Deoria, and Mau. Sample collection for UP was undertaken during 2021 (post-COVID), which showed a minimal decrease in the concentration of arsenic in a few regions in comparison to the values of 2015. Although system parameters and their interplay suggest, mobilization could have been more. Still, it was limited due to the restrictions on anthropogenic activities such as the chemical industry, nonferrous metal mining, smelting, and pesticide production during the lockdown imposed by the Indian government. The observed change was minimal because of the continued groundwater abstraction and agricultural practices during the COVID period. Figure 7 has four parts, and each part represents four sections: (i) predicted arsenic hazard with the data points in the selected state boundaries; (ii) predicted arsenic hazard and the data points with district boundaries for the selected districts; (iii) the percentage of arsenic occurrence; and (iv) a scatterplot of the predicted hazard index values and actual arsenic concentrations with correlation coefficient (r). The correlation coefficient (r) for all four cases is in the range of 0.8–0.9 (quite close to 1). It follows that the actual arsenic concentrations show a positive linear relationship with the predicted arsenic hazard.
Hence, the study models demonstrated perfect accuracy, proving that the trained models and the physically monitored field values demonstrated a substantial agreement. Therefore, the generated hazard maps can help stakeholders plan an efficient mitigation strategy for the elevated arsenic contamination in the basin.

4. Conclusions

This study endeavoured to determine groundwater arsenic hazard using the AHP and FR models for the entire Ganga basin. A total of 15 variables were selected to reflect on the possible occurrence of arsenic in groundwater on the premise of their association with it. The AHP considered the weightage of each covariate, while the FR used the relationship of arsenic with variables. Most of the regions around the Ganga River showed high arsenic contamination. The results showed that 35% and 37% of the basin area are under high arsenic hazard using the AHP and FR models. Both hierarchy and arsenic frequency statistics methods provided matching results with minimal differences.
The validation of the output demonstrated that the AUROC curve value ranged from 0.7 to 0.8 for both the models, demonstrating 70–80% of the accuracy for both training and testing datasets. This observation shows the robust nature of the models. Yet another accuracy assessment conducted by comparing the predicted groundwater arsenic hazard values with the primary arsenic data (2019, 2021) collected for the districts of Bihar and UP indicated that the hazard predictions were realistic and represented the actual field situation. This study implies that the generated groundwater arsenic hazard maps may help stakeholders plan an efficient mitigation strategy for arsenic management in the Ganga basin.
Furthermore, the present study demonstrates the major contribution of geogenic factors toward the leaching of arsenic into the groundwater. However, organics from built-up and agricultural areas, which cannot be neglected, attributed to contaminating the groundwater. It is observed that the arsenic-contaminated groundwater is near-neutral and confined to the alluvium plains with loads of silicate minerals, iron, and bicarbonates (HCO3). Arsenic shows a strong positive and negative correlation with physicochemical groundwater quality parameters, which either release or bind arsenic species. Hence, it plays a vital role in either increasing or preventing arsenic contamination in groundwater. Therefore, the presence of arsenic in the basin largely depends upon the geomorphology, land-use patterns, and existing groundwater quality.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/w14152440/s1, The supporting information was used to generate the results. Table S1: Weightage Evaluation for Analytical Hierarchy Process (AHP) Model; Table S2: Frequency Ratio (FR) And Prediction Rate (PR).

Author Contributions

The work was conceptualized by S.D. under the supervision of H.J. The framework to run the model was prepared by S.D. and reviewed by H.J. Data analysis, modelling, and validation were performed by S.D. in ArcMap v 10.6.1. All the outcomes were reviewed and improved by H.J. Collectively S.D. and H.J. have written the manuscript. All the authors have discussed and finalized the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

The financial support from Department of Science and Technology (DST) for providing Inspire fellowship with the grant number (7069-111-044-428) to the first author is deeply appreciated. This work was supported by the Department of Science and Technology (DST, India) (vide no. DST/TM/INDO-UK/2K17/55(C) & 55(G))-Newton Bhabha-Natural Environmental Research Council (NERC, UK (NE/R003386/1) through providing financial assistance for field visits and analysis related costs.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Sources of all the secondary data used are already provided in the manuscript. Currently, authors are not in position to share the primary field data but can be available on request from the corresponding author.

Acknowledgments

The authors are thankful to CGWB, Geological Survey of India, United States Geological Sciences, Indian Water Resource Information System, and Ministry of Jal Shakti for providing secondary data for smooth research work. We are grateful to the Department of Hydrology and Institute Instrumentation Center of IIT Roorkee for providing the equipment for water quality analysis and specific services of ICPMS for primary field data analysis, respectively. Maps throughout the paper were created in ArcGIS® and ArcMapTM v 10.6.1 software, intellectual property of Esri, and used under license with Copyright © Esri.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Central Ground Water Board (CGWB). Arsenic Hotspots in Ground Water in India. MoWR, G.o.I. 2015. Available online: http://cgwb.gov.in/WQ/ARSENIC.pdf (accessed on 4 August 2019).
  2. Alarcón-Herrera, M.T.; Gutiérrez, M. Geogenic Arsenic in Groundwater: Challenges, Gaps and Future Directions. Curr. Opin. Environ. Sci. Health 2022, 27, 100349. [Google Scholar] [CrossRef]
  3. Fuoco, I.; Marini, L.; De Rosa, R.; Figoli, A.; Gabriele, B.; Apollaro, C. Use of reaction path modelling to investigate the evolution of water chemistry in shallow to deep crystalline aquifers with a special focus on fluoride. Sci. Total Environ. 2022, 830, 154566. [Google Scholar] [CrossRef] [PubMed]
  4. National Research Council (US) Committee on Medical and Biological Effects of Environmental Pollutants. Arsenic: Medical and Biologic Effects of Environmental Pollutants; National Academies Press: Washington, DC, USA, 1977. Available online: https://nap.nationalacademies.org/read/9003/chapter/1 (accessed on 24 December 2021).
  5. Shaji, E.; Santosh, M.; Sarath, K.V.; Prakash, P.; Deepchand, V.; Divya, B.V. Arsenic contamination of groundwater: A global synopsis with focus on the Indian Peninsula. Geosci. Front. 2021, 12, 101079. [Google Scholar] [CrossRef]
  6. Fuoco, I.; De Rosa, R.; Barca, D.; Figoli, A.; Gabriele, B.; Apollaro, C. Arsenic polluted waters: Application of geochemical modelling as a tool to understand the release and fate of the pollutant in crystalline aquifers. J. Environ. Manag. 2022, 301, 113796. [Google Scholar] [CrossRef] [PubMed]
  7. Polya, D.A.; Lawson, M. Geogenic and anthropogenic arsenic hazards in groundwater and soils: Distribution, nature, origin and human exposure routes. In Arsenic: Exposure Sources, Health Risks and Mechanisms of Toxicity; States, J.C., Ed.; John Wiley & Sons: Hoboken, NJ, USA, 2017; pp. 23–60. [Google Scholar] [CrossRef]
  8. Polya, D.A.; Middleton, D.R.S. Arsenic in drinking water: Sources & human exposure routes. In Best Practice Guide for the Control of Arsenic in Drinking Water; Bhattacharya, P., Polya, D.A., Jovanovic., D., Eds.; IWA Publishing: London, UK, 2017; pp. 1–14. [Google Scholar]
  9. Islam, F.S.; Gault, A.G.; Boothman, C.; Polya, D.A.; Charnock, J.M.; Chatterjee, D.; Lloyd, J.R. Role of Metal Reducing Bacteria in Arsenic Release in Bengal Delta Sediments. Nature 2004, 430, 68–71. [Google Scholar] [CrossRef]
  10. Jang, Y.; Somanna, Y.; Kim, H. Source, Distribution, Toxicity and Remediation of Arsenic in the Environment–A Review. Int. J. Appl. Environ. Sci. 2016, 11, 559–581. [Google Scholar]
  11. Matschullat, J. Arsenic in the Geosphere–A Review. Sci. Total Environ. 2000, 249, 297–312. [Google Scholar] [CrossRef]
  12. Nordstrom, D.K. Worldwide occurrence of arsenic in groundwater. Science 2002, 296, 2143–2145. [Google Scholar] [CrossRef]
  13. Hindmarsh, J.T.; McCurdy, R.F.; Savory, J. Clinical and Environmental Aspects of Arsenic Toxicity. Crit. Rev. Clin. Lab. Sci. 1986, 23, 315–347. [Google Scholar] [CrossRef]
  14. Squibb, K.S.; Fowler, B.A. The toxicity of arsenic and its compounds. In Biological and Environmental Effects of Arsenic; Fowler, B.A., Ed.; Elsevier Science Publishers, B.V.: Amsterdam, The Netherlands, 1983; pp. 233–269. [Google Scholar]
  15. Chatterjee, S.; Moogoui, R.; Gupta, D.K. Arsenic: Source, Occurrence, Cycle, and Detection. In Arsenic Contamination in the Environment; Gupta, D., Chatterjee, S., Eds.; Springer Nature: Cham, Switzerland, 2017; pp. 13–35. [Google Scholar] [CrossRef]
  16. Lloyd, J.; Oremland, R. Microbial Transformations of Arsenic in the Environment: From Soda Lakes to Aquifers. Elements 2006, 2, 85–90. [Google Scholar] [CrossRef]
  17. Páez-Espino, D.; Tamames, J.; de Lorenzo, V.; Cánovas, D. Microbial responses to environmental arsenic. Biometals 2009, 22, 117–130. [Google Scholar] [CrossRef] [PubMed]
  18. Kim, K.W.; Chanpiwat, P.; Hanh, H.T.; Phan, K.; Sthiannopkao, S. Arsenic geochemistry of groundwater in Southeast Asia. Front. Med. 2011, 5, 420–433. [Google Scholar] [CrossRef] [PubMed]
  19. WHO. Arsenic. In WHO Air Quality Guidelines for Europe, 2nd ed.; WHO Regional Publications: Copenhagen, Denmark, 2000. [Google Scholar]
  20. BIS. IS 10500: 2012 Drinking Water—Specification, 2nd ed.; BIS Publication Unit: New Delhi, India, 2012. [Google Scholar]
  21. Bindal, S.; Singh, C.K. Predicting groundwater arsenic contamination: Regions at risk in highest populated state of India. Water Res. 2019, 159, 65–76. [Google Scholar] [CrossRef] [PubMed]
  22. Chakraborty, M.; Sarkar, S.; Mukherjee, A.; Shamsudduha, M.; Ahmed, K.M.; Bhattacharya, A.; Mitra, A. Modeling regional-scale groundwater arsenic hazard in the transboundary Ganges River Delta, India and Bangladesh: Infusing physically-based model with machine learning. Sci. Total Environ. 2020, 748, 141107. [Google Scholar] [CrossRef] [PubMed]
  23. Jangle, N.; Sharma, V.; Dror, D.M. Statistical geospatial modelling of arsenic concentration in Vaishali District of Bihar, India. Sustain. Water Resour. Manag. 2016, 2, 285–295. [Google Scholar] [CrossRef] [Green Version]
  24. Mukherjee, A.; Sarkar, S.; Chakraborty, M.; Duttagupta, S.; Bhattacharya, A.; Saha, D.; Bhattacharya, P.; Mitra, A.; Gupta, S. Occurrence, predictors and hazards of elevated groundwater arsenic across India through field observations and regional-scale AI-based modeling. Sci. Total Environ. 2020, 759, 143511. [Google Scholar] [CrossRef]
  25. Nath, B.; Chowdhury, R.; Ni-Meister, W.; Mahanta, C. Predicting the distribution of arsenic in groundwater by a geospatial machine learning technique in the two most affected districts of Assam, India: The public health implications. GeoHealth 2022, 6, e2021GH000585. [Google Scholar] [CrossRef]
  26. Podgorski, J.; Wu, R.; Chakravorty, B.; Polya, D.A. Groundwater Arsenic Distribution in India by Machine Learning Geospatial Modeling. Int. J. Environ. Res. Public Health 2020, 17, 7119. [Google Scholar] [CrossRef]
  27. Podgorski, J.; Berg, M. Global threat of arsenic in groundwater. Science 2020, 368, 845–850. [Google Scholar] [CrossRef]
  28. Wu, R.; Podgorski, J.; Berg, M.; Polya, D.A. Geostatistical model of the spatial distribution of arsenic in groundwaters in Gujarat State, India. Environ. Geochem. Health 2021, 43, 2649–2664. [Google Scholar] [CrossRef]
  29. Chakraborti, D.; Singh, S.K.; Rahman, M.M.; Dutta, R.N.; Mukherjee, S.C.; Pati, S.; Kar, P.B. Groundwater Arsenic Contamination in the Ganga River Basin: A Future Health Danger. Int. J. Environ. Res. Public Health 2018, 15, 180. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  30. Amarasinghe, U.A.; Muthuwatta, L.; Smakhtin, V.; Surinaidu, L.; Natarajan, R.; Chinnasamy, P.; Kakumanu, K.R.; Prathapar, S.A.; Jain, S.K.; Ghosh, N.C.; et al. Reviving the Ganges water machine: Potential and challenges to meet increasing water demand in the Ganges River Basin. In IWMI Research Report 167; IWMI: Colombo, Sri Lanka, 2016; p. 42. [Google Scholar] [CrossRef]
  31. Sharma, B.; Amarasinghe, U.; Xueliang, C.; de Condappa, D.; Shah, T.; Mukherji, A.; Smakhtin, V. The Indus and the Ganges: River basins under extreme pressure. Water Int. 2010, 35, 493–521. [Google Scholar] [CrossRef]
  32. Kaushal, N.; Kansal, M.L. Overview of water allocation practices in Uttar Pradesh and Uttarakhand with a specific reference to future demands. In report of SAWAS 2011, 2. Available online: http://saciwaters.org/sawasjournal/Full-SAWAS%202(2),%202011.pdf (accessed on 11 February 2022).
  33. Sinha, R.; Tandon, S.K.; Gibling, M.R. Shallow sub-surface stratigraphy of the Ganga basin, Himalayan foreland: Present status and future perspectives. Quat. Int. 2010, 227, 81–86. [Google Scholar] [CrossRef]
  34. Bhukosh-Geological Survey of India. Available online: https://bhukosh.gsi.gov.in/Bhukosh/Public (accessed on 26 November 2020).
  35. Dhamija, S.; Joshi, H. Vulnerability assessment of arsenic in groundwater and its occurrence in districts of Bihar, India. In Proceedings of the National Seminar on Recent Advances in Geospatial Technology & Applications, IIRS, Dehradun, India, 2 March 2020; pp. 158–164. Available online: https://www.iirs.gov.in/national-seminar (accessed on 20 August 2020).
  36. Saranya, T.; Saravanan, S. Assessment of groundwater vulnerability using analytical hierarchy process and evidential belief function with DRASTIC parameters, Cuddalore, India. Int. J. Environ. Sci. Technol. 2022, 9, 20. [Google Scholar] [CrossRef]
  37. Ahmadi, H.; Kaya, O.A.; Babadagi, E.; Savas, T.; Pekkan, E. GIS-based groundwater potentiality mapping using AHP and FR models in central antalya, Turkey. Environ. Sci. Proc. 2021, 5, 11. [Google Scholar] [CrossRef]
  38. Hasanuzzaman, M.; Adhikary, P.P.; Bera, B.; Shit, P.K. Flood Vulnerability Assessment Using AHP and Frequency Ratio Techniques. In Spatial Modelling of Flood Risk and Flood Hazards; Pradhan, B., Shit, P.K., Bhunia, G.S., Adhikary, P.P., Pourghasemi, H.R., Eds.; Springer: Cham, Switzerland, 2022; pp. 91–104. [Google Scholar] [CrossRef]
  39. Esri. Available online: https://www.esri.com/en-us/what-is-gis/overview (accessed on 15 December 2020).
  40. India-WRIS. Available online: https://indiawris.gov.in/wris/#/dataAvailability (accessed on 15 December 2020).
  41. USGS. Available online: https://earthexplorer.usgs.gov (accessed on 12 August 2020).
  42. Geological Survey of India. Available online: https://www.gsi.gov.in/webcenter/portal (accessed on 30 August 2020).
  43. CGWB. Available online: http://cgwb.gov.in/wqreports.html (accessed on 4 August 2019).
  44. Food and Agriculture Organization of the United Nations. Available online: http://www.fao.org/soils-portal/data-hub (accessed on 16 January 2021).
  45. CHRS Data Portal. Available online: https://chrsdata.eng.uci.edu (accessed on 30 January 2021).
  46. Apollaro, C.; Di Curzio, D.; Fuoco, I.; Buccianti, A.; Dinelli, E.; Vespasiano, G.; Castrignanò, A.; Rusi, S.; Barca, D.; Figoli, A.; et al. A multivariate non-parametric approach for estimating probability of exceeding the local natural background level of arsenic in the aquifers of Calabria region (Southern Italy). Sci. Total Environ. 2022, 806, 150345. [Google Scholar] [CrossRef]
  47. Saaty, T.L. Decision Making with the Analytic Hierarchy Process. Int. J. Serv. Sci. 2008, 1, 83. [Google Scholar] [CrossRef] [Green Version]
  48. Lee, A.H.I.; Wen, C.C.; Ching, J.C. A Fuzzy AHP and BSC Approach for Evaluating Performance of IT Department in the Manufacturing Industry in Taiwan. Expert Syst. Appl. 2008, 34, 96–107. [Google Scholar] [CrossRef]
  49. Mann, S.H. An evaluation of the eigenvalue approach. Fuzzy Sets Syst. 1990, 35, 295–301. [Google Scholar] [CrossRef]
  50. Bonham-Carter, G.F. Geographic Information Systems for Geoscientists: Modelling with GIS; Love Printing Services: Ottawa, ON, Canada, 1994; Volume 13. [Google Scholar] [CrossRef]
  51. Guru, B.; Seshan, K.; Bera, S. Frequency Ratio Model for Groundwater Potential Mapping and Its Sustainable Management in Cold Desert, India. J. King Saud Univ. Sci. 2017, 29, 333–347. [Google Scholar] [CrossRef] [Green Version]
  52. Das, S. Comparison among Influencing Factor, Frequency Ratio, and Analytical Hierarchy Process Techniques for Groundwater Potential Zonation in Vaitarna Basin, Maharashtra, India. Groundw. Sustain. Dev. 2019, 8, 617–629. [Google Scholar] [CrossRef]
  53. Lee, S.; Pradhan, B. Landslide hazard mapping at Selangorm, Malaysia using frequency ratio and logistic regression models. Landslides 2007, 4, 33–41. [Google Scholar] [CrossRef]
  54. Althuwaynee, O.F.; Pradhan, B.; Park, H.; Lee, J. A novel ensemble bivariate statistical evidential belief function with knowledge-based analytical hierarchy process and multivariate statistical logistic regression for landslide susceptibility mapping. Catena 2014, 114, 21–36. [Google Scholar] [CrossRef]
  55. Nampak, H.; Pradhan, B.; Manap, M.A. Application of GIS Based Data Driven Evidential Belief Function Model to Predict Groundwater Potential Zonation. J. Hydrol. 2014, 513, 283–300. [Google Scholar] [CrossRef]
  56. Hadžić, E.; Lazović, N.; Mulaomerović-Šeta, A. The Importance of Groundwater Vulnerability Maps in the Protection of Groundwater Sources. Key Study: Sarajevsko Polje. Procedia Environ. Sci. 2015, 25, 104–111. [Google Scholar] [CrossRef] [Green Version]
  57. Fawcett, T. An introduction to ROC analysis. Pattern Recognit. Lett. 2006, 27, 861–874. [Google Scholar] [CrossRef]
  58. Swets, J.A. Measuring the accuracy of diagnostic systems. Science 1988, 240, 1285–1293. [Google Scholar] [CrossRef] [Green Version]
  59. Franson, M.A.H. Standard Methods for the Examination of Water and Wastewater; Clesceri, L.S., Greenberg, A.E., Eaton, A.D., Eds.; American Public Health Association: Washington, DC, USA, 1999; pp. 1027–1033. [Google Scholar]
  60. Chakroborti, D.; Das, B.; Murrill, T.M. Examining India’s Groundwater Quality Management. Environ. Sci. Technol. 2011, 45, 27–33. [Google Scholar] [CrossRef]
  61. Saha, D.; Sreehari, S.; Dwivedi, S.N.; Bhartariya, K.G. Evaluation of hydrogeochemical processes in arsenic contaminated alluvial aquifers in parts of Mid Ganga Basin, Bihar, Eastern India. Environ. Earth Sci. 2009, 61, 799–811. [Google Scholar] [CrossRef]
  62. Kinniburgh, D.G.; Smedley, P.L.; Davies, J.; Milne, C.J.; Gaus, I.; Tradfford, J.M.; Burden, S.; Huq, S.M.I.; Ahmad, N.; Ahmed, K.M. The scale and causes of the groundwater arsenic problem in Bangladesh. In Arsenic in Ground Water; Welch, A.H., Stollenwerk, K.G., Eds.; Springer: Boston, MA, USA, 2003; pp. 211–257. [Google Scholar] [CrossRef] [Green Version]
  63. Herath, I.; Vithanage, M.; Bundschuh, J.; Maity, J.P.; Bhattacharya, P. Natural Arsenic in Global Groundwaters: Distribution and Geochemical Triggers for Mobilization. Curr. Pollut. Rep. 2016, 2, 68–89. [Google Scholar] [CrossRef] [Green Version]
  64. McArthur, J.M.; Ravenscroft, P.; Banerjee, D.M.; Milsom, J.; Hudson-Edwards, K.A.; Sengupta, S.; Bristow, C.; Sarkar, A.; Tonkin, S.; Purohit, R. How paleosols influence groundwater flow and arsenic pollution: A model from the Bengal Basin and its worldwide implication. Water Resour. Res. 2008, 44, 11. [Google Scholar] [CrossRef]
  65. Lynch, S.D.; Reynders, A.G.; Schulze, R.E. Preparing input data for a national-scale groundwater vulnerability map of southern Africa. Water SA 1994, 20, 239–246. [Google Scholar]
  66. Chakraborty, M.; Mukherjee, A.; Ahmed, K.M. A Review of Groundwater Arsenic in the Bengal Basin, Bangladesh and India: From Source to Sink. Curr. Pollution Rep. 2015, 1, 220–247. [Google Scholar] [CrossRef] [Green Version]
  67. Bhattacharya, P.; Chatterjee, D.; Jacks, G. Occurrence of Arsenic-contaminated Groundwater in Alluvial Aquifers from Delta Plains, Eastern India: Options for Safe Drinking Water Supply. Int. J. Water Resour. Dev. 1997, 13, 79–92. [Google Scholar] [CrossRef]
  68. Das, D.; Chatterjee, A.; Mandal, B.K.; Samanta, G.; Chakraborti, D.; Chanda, B. Arsenic in ground water in six districts of West Bengal, India. Environ. Geochem. Health 1995, 18, 5–15. [Google Scholar] [CrossRef]
  69. Nickson, R.; McArthur, J.; Burgess, W.; Ahmed, K.M.; Ravenscroft, P.; Rahman, M. Arsenic poisoning of Bangladesh groundwater. Nature 1998, 395, 338. [Google Scholar] [CrossRef] [PubMed]
  70. Acharyya, S.K. Arsenic Levels in Groundwater from Quaternary Alluvium in the Ganga Plain and the Bengal Basin, Indian Subcontinent: Insights into Influence of Stratigraphy. GR 2005, 8, 55–66. [Google Scholar] [CrossRef]
  71. Vaughn, P. A Basic Study in Groundwater and the Hydrogeologic characteristics of Principal Aquifers in The United States. Master’s Thesis, University of Florida, Gainesville, FL, USA, July 2015. [Google Scholar]
  72. Wang, S.; Mulligan, C.N. Occurrence of Arsenic Contamination in Canada: Sources, Behaviour and Distribution. Sci. Total Environ. 2006, 366, 701–721. [Google Scholar] [CrossRef]
  73. Mandal, B.K.; Chowdhury, T.R.; Samanta, G.; Basu, G.K.; Chowdhury, P.P.; Chanda, C.R.; Lodh, D.; Karan, N.K.; Dhara, R.K.; Tamili, D.K.; et al. Arsenic in groundwater in seven districts of West Bengal, India–The biggest arsenic calamity in the world. Curr. Sci. 1996, 70, 976–986. [Google Scholar]
  74. Pandey, V.C.; Singh, J.S.; Singh, R.P.; Wingh, N.; Yunus, M. Arsenic hazards in coal fly ash and its fate in Indian scenario. Resour. Conserv. Recycl. 2011, 55, 819–835. [Google Scholar] [CrossRef]
  75. Mallick, S.; Rajagopal, N.R. Groundwater development in the arsenic-affected alluvial belt of West Bengal—Some questions. Curr. Sci. 1996, 70, 956–958. Available online: https://www.jstor.org/stable/24111630 (accessed on 10 March 2022).
  76. Anawar, H.M.; Tareq, S.M.; Ahmed, G. Is organic matter a source or redox driver or both for arsenic release in groundwater? Phys. Chem. Earth 2013, 58–60, 49–56. [Google Scholar] [CrossRef]
  77. Savarimuthu, X.; Hira-Smith, M.M.; Yuan, Y.; von Ehrenstein, O.S.; Das, S.; Ghosh, N.; Mazumder, D.N.; Smith, A.H. Seasonal variation of arsenic concentrations in tube wells in west Bengal, India. J. Health Popul. Nutr. 2006, 24, 277–281. [Google Scholar]
  78. Ghosh, T.; Kanchan, R. Aquifer Vulnerability Assessment in the Bengal Alluvial Tract, India, Using GIS Based DRASTIC Model. MESE 2016, 2, 1–13. [Google Scholar] [CrossRef] [Green Version]
  79. Rodríguez, R.; Ramos, J.; Armienta, A. Groundwater arsenic variations: The role of local geology and rainfall. Appl. Geochem. 2004, 19, 245–250. [Google Scholar] [CrossRef]
  80. Planer-Friedrich, B.; Härtig, C.; Lissner, H.; Steinborn, J.; Süß, E.; Qumrul Hassan, M.; Zahid, A.; Alam, M.; Merkel, B. Organic carbon mobilization in a Bangladesh aquifer explained by seasonal monsoon-driven storativity changes. App. Geochem. 2012, 27, 2324–2334. [Google Scholar] [CrossRef]
  81. Pradeep, K.; Nepolian, M.; Anandhan, P.; Chandran; Kaviyarasan, R.; Prasanna, M.V.; Chidambaram, S. A study on variation in dissolved silica concentration in groundwater of hard rock aquifers in Southeast coast of India. In Proceedings of the 10th Curtin University Technology Science and Engineering International Conference (CUTSE2015), Miri, Malaysia, 6–8 November 2015. [Google Scholar] [CrossRef]
  82. Alam, M.S.; Wu, Y.; Cheng, T. Silicate Minerals as a Source of Arsenic Contamination in Groundwater. Wat. Air Soil Poll. 2014, 225, 2201. [Google Scholar] [CrossRef]
  83. Alam, M.O.; Shaikh, W.A.; Chakraborty, S.; Avishek, K.; Bhattacharya, T. Groundwater Arsenic Contamination and Potential Health Risk Assessment of Gangetic Plains of Jharkhand, India. Expos. Health 2016, 8, 125–142. [Google Scholar] [CrossRef]
  84. Ravindra, K.; Garg, V.K. Hydro-chemical survey of groundwater of Hisar City and assessment of defluoridation methods used in India. Environ. Monit. Assess. 2007, 132, 33–43. [Google Scholar] [CrossRef]
  85. Bothe, J.; James, V.; Brown, P.W. Arsenic immobilization by calcium arsenate formation. Environ. Sci. Technol. 1999, 33, 3806–3811. [Google Scholar] [CrossRef]
  86. Raposo, J.C.; Zuloaga, O.; Olazabal, M.A.; Madariaga, J.M. Study of the precipitation equilibria of arsenate anion with calcium and magnesium in sodium perchlorate at 25 °C. Appl. Geochem. 2004, 19, 855–862. [Google Scholar] [CrossRef]
  87. Smith, E.; Naidu, R.; Alston, A.M. Chemistry of inorganic arsenic in soils: II. Effect of phosphorus, sodium, and calcium on arsenic sorption. J. Environ. Qual. 2002, 31, 557–563. [Google Scholar] [CrossRef] [PubMed]
  88. Wilkie, J.A.; Hering, J.G. Adsorption of arsenic onto hydrous ferric oxide: Effects of adsorbate/adsorbent ratios and co-occurring solutes. Colloids Surf. A Physicochem. Eng. Asp. 1996, 107, 97–110. [Google Scholar] [CrossRef]
  89. Lien, H.; Wilkin, R. High-level arsenite removal from groundwater by zero-valent iron. Chemosphere 2005, 59, 377–386. [Google Scholar] [CrossRef] [PubMed]
  90. Smedley, P.L.; Kinniburgh, D.G. A review of the source, behaviour and distribution of arsenic in natural waters. Appl. Geochem. 2002, 17, 517–568. [Google Scholar] [CrossRef] [Green Version]
  91. Ohno, K.; Furukawa, A.; Hayashi, K.; Kamei, T.; Magara, Y. Arsenic contamination of groundwater in Nawabganj, Bangladesh, focusing on the relationship with other metals and ions. Water Sci. Technol. 2005, 52, 87–94. [Google Scholar] [CrossRef] [PubMed]
  92. Maliva, R.G. Anthropogenic Aquifer Recharge; Series no. 5; Springer Nature: Cham, Switzerland, 2020. [Google Scholar] [CrossRef]
  93. Ahmann, D.; Roberts, A.L.; Krumholz, L.R.; Morel, F.M. Microbe grows by reducing arsenic. Nature 1994, 371, 750. [Google Scholar] [CrossRef] [PubMed]
  94. Bhattacharya, P.; Mukherjee, A.; Mukherjee, A.B. Arsenic in Groundwater of India. In Encyclopedia of Environmental Health; Nriagu, J.O., Ed.; Elsevier: Amsterdam, The Netherland, 2011; pp. 150–164. [Google Scholar] [CrossRef]
  95. Gulz, P.A.; Gupta, S.; Schulin, R. Arsenic accumulation of common plants from contaminated soils. Plant Soil 2005, 272, 337–347. [Google Scholar] [CrossRef]
  96. Smith, E.; Naidu, R.; Alston, A.M. Arsenic in the soil environment: A review. Adv. Agron. 1998, 64, 149–195. [Google Scholar]
  97. Jenks, G.F. The data model concept in statistical mapping. Int. Year Book Cartogr. 1967, 7, 186–190. [Google Scholar]
  98. Wei, X.; Zhang, L.; Luo, J.; Liu, D. A hybrid framework integrating physical model and convolutional neural network for regional landslide susceptibility mapping. Nat. Hazards 2021, 109, 471–497. [Google Scholar] [CrossRef]
  99. Pradhan, B. A comparative study on the predictive ability of the decision tree, support vector machine and neuro-fuzzy models in landslide susceptibility mapping using GIS. Comput. Geosci. 2013, 51, 350–365. [Google Scholar] [CrossRef]
  100. Pradhan, B. Groundwater potential zonation for basaltic watersheds using satellite remote sensing data and GIS techniques. Cent. Eur. J. Geosci. 2009, 1, 120–129. [Google Scholar] [CrossRef]
  101. Naghibi, S.A.; Pourghasemi, H.R.; Pourtaghi, Z.S. Groundwater qanat potential mapping using frequency ratio and Shannon’s entropy models in the Moghan watershed, Iran. Earth Sci. Inform. 2015, 8, 171–186. [Google Scholar] [CrossRef]
  102. Yesilnacar, E.K. The Application of Computational Intelligence to Landslide Susceptibility Mapping in Turkey. Ph.D. Thesis, University of Melbourne, Parkville, Australia, 2005; 423p. [Google Scholar]
  103. Yesilnacar, E.; Topal, T. Landslide susceptibility mapping: A comparison of logistic regression and neural networks methods in a medium scale study, Hendek region (Turkey). Eng. Geol. 2005, 79, 251–266. [Google Scholar] [CrossRef]
  104. Muavhi, N.; Thamaga, K.H.; Mutoti, M.I. Mapping groundwater potential zones using relative frequency ratio, analytic hierarchy process and their hybrid models: Case of Nzhelele-Makhado area in South Africa. Geocarto Int. 2021, 1–20. [Google Scholar] [CrossRef]
Figure 1. Study area—Ganga Basin, India [34].
Figure 1. Study area—Ganga Basin, India [34].
Water 14 02440 g001
Figure 2. Schematic flowchart of the methodology of arsenic occurrence, assessment, and mapping.
Figure 2. Schematic flowchart of the methodology of arsenic occurrence, assessment, and mapping.
Water 14 02440 g002
Figure 3. Thematic layers and arsenic occurrence percentages w.r.t.: (a) DTW; (b) slope; (c) geomorphology; (d) types of aquifers; (e) soil types; (f) LULC; (g) rainfall; (h) groundwater abstraction; (i) dissolved silica; (j) bicarbonate; (k) iron; (l) EC; (m) hardness; (n) sulphate; and (o) testing arsenic datasets of the study area.
Figure 3. Thematic layers and arsenic occurrence percentages w.r.t.: (a) DTW; (b) slope; (c) geomorphology; (d) types of aquifers; (e) soil types; (f) LULC; (g) rainfall; (h) groundwater abstraction; (i) dissolved silica; (j) bicarbonate; (k) iron; (l) EC; (m) hardness; (n) sulphate; and (o) testing arsenic datasets of the study area.
Water 14 02440 g003aWater 14 02440 g003bWater 14 02440 g003cWater 14 02440 g003d
Figure 4. Groundwater arsenic hazard map for the Ganga basin using (a) AHP, (b) FR.
Figure 4. Groundwater arsenic hazard map for the Ganga basin using (a) AHP, (b) FR.
Water 14 02440 g004
Figure 5. Bar graph of groundwater arsenic prediction by geostatistical methods.
Figure 5. Bar graph of groundwater arsenic prediction by geostatistical methods.
Water 14 02440 g005
Figure 6. (a) AUROC validation curve with training datasets for the AHP and FR methods; (b) AUROC validation curve with testing datasets for the AHP and FR methods.
Figure 6. (a) AUROC validation curve with training datasets for the AHP and FR methods; (b) AUROC validation curve with testing datasets for the AHP and FR methods.
Water 14 02440 g006
Figure 7. Validation through actual field data for: (a) AHP (Bihar); (b) FR (Bihar); (c) AHP (UP); (d) FR (UP). Each part represents four sections: (i) predicted arsenic hazard with the data points in the selected state boundaries; (ii) predicted arsenic hazard and the data points with the district boundaries for the selected districts; (iii) the percentage of arsenic occurrence; and (iv) a scatterplot of the predicted hazard index values and actual arsenic concentrations.
Figure 7. Validation through actual field data for: (a) AHP (Bihar); (b) FR (Bihar); (c) AHP (UP); (d) FR (UP). Each part represents four sections: (i) predicted arsenic hazard with the data points in the selected state boundaries; (ii) predicted arsenic hazard and the data points with the district boundaries for the selected districts; (iii) the percentage of arsenic occurrence; and (iv) a scatterplot of the predicted hazard index values and actual arsenic concentrations.
Water 14 02440 g007
Table 1. Selected variables and the data sources.
Table 1. Selected variables and the data sources.
S.No.VariablesData SourceFormatYear
1.Depth to Water level (DTW)Water Resource Information System (WRIS) [40]Tabular2015
2.SlopeUnited States Geological Survey (USGS) [41]Raster2015
3.GeomorphologyGeological Survey of India (GSI) [42]Map2007
4.Types of aquifersCentral Ground Water Board (CGWB) [43]Tabular2010
5.SoilThe Food and Agriculture Organization of the United Nations (FAO) [44] Raster2007
6.Land Use Land Cover (LULC)United States Geological Survey (USGS) [41]Landsat Imagery2015
7.RainfallCentre for Hydrometeorology and Remote Sensing (CHRS) [45]Raster2015
8.Groundwater AbstractionCentral Ground Water Board (CGWB) [43]Tabular2013
9.Groundwater Quality ParametersCentral Ground Water Board (CGWB) [43]Tabular2015
Table 2. Saaty ratio scale for pair-wise comparison.
Table 2. Saaty ratio scale for pair-wise comparison.
Intensity of ImportanceDefinitionExplanation
1Equal importanceTwo activities contribute equally to the objective
3Weak importance of one over anotherExperience and judgment slightly favour one activity over another
5Essential or strong importanceExperience and judgment strongly favour one activity over another
7Demonstrated importanceActivity is strongly favoured, and its dominance is demonstrated in practice
9Absolute importanceThe evidence favouring one activity over another is of the highest possible order of affirmation
2,4,6,8Intermediate values between the two adjacent judgementsWhen compromise is needed
Table 3. Random consistency index.
Table 3. Random consistency index.
Number
of Rows (n)
123456789101112131415
RCI000.580.91.121.241.321.411.451.491.511.481.561.571.59
Table 4. Correlation of arsenic with groundwater quality parameters.
Table 4. Correlation of arsenic with groundwater quality parameters.
pHEC THCa2+Mg2+Na+K+CO32−HCO3ClNO3SO42−FPO43−SiO4Fe As
pH1
EC−0.0241
TH−0.0340.7021
Ca2+−0.0260.5380.6971
Mg2+0.0330.6330.8910.3561
Na+−0.0200.8670.5270.3850.4821
K+−0.0100.1990.1700.0850.1760.1041
CO32−0.6780.0830.008−0.0560.0880.121−0.0061
HCO3−0.0430.5620.4610.3860.4060.5360.2310.0331
Cl−0.0170.8620.6930.5150.6160.8880.1390.0630.3561
NO3−0.0140.4190.3490.3520.3270.3720.2610.1090.2690.3121
SO42−−0.0180.6490.5300.4120.4900.6440.1430.0580.3480.5000.3511
F−0.0160.3510.1270.0170.1570.4720.0120.0440.3130.3480.0970.2141
PO43−0.659−0.032−0.046−0.037−0.002−0.0220.0620.472−0.017−0.025−0.043−0.046−0.0501
SiO4−0.056−0.0500.0210.018−0.005−0.0890.053−0.1070.057−0.082−0.124−0.066−0.055−0.0321
Fe−0.004−0.066−0.018−0.1100.040−0.046−0.015−0.031−0.068−0.036−0.038−0.097−0.0520.070−0.1281
As−0.011−0.449−0.3340.0220.019−0.0130.006−0.0040.249−0.001−0.002−0.392−0.0050.0110.5440.2941
Table 5. Arsenic prediction by selected geostatistical methods.
Table 5. Arsenic prediction by selected geostatistical methods.
S.No.Hazard LikelihoodArea (AHP)% AreaArea (FR)% Area
1.Very low76,753962,3788
2.Low193,60224243,58130
3.Moderate253,28431194,98824
4.High287,23335298,55337
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Dhamija, S.; Joshi, H. Prediction of Groundwater Arsenic Hazard Employing Geostatistical Modelling for the Ganga Basin, India. Water 2022, 14, 2440. https://doi.org/10.3390/w14152440

AMA Style

Dhamija S, Joshi H. Prediction of Groundwater Arsenic Hazard Employing Geostatistical Modelling for the Ganga Basin, India. Water. 2022; 14(15):2440. https://doi.org/10.3390/w14152440

Chicago/Turabian Style

Dhamija, Sana, and Himanshu Joshi. 2022. "Prediction of Groundwater Arsenic Hazard Employing Geostatistical Modelling for the Ganga Basin, India" Water 14, no. 15: 2440. https://doi.org/10.3390/w14152440

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop