Human health risk and hydro-geochemical appraisal of groundwater in the southwest part of Bangladesh using GIS, water quality indices, and multivariate statistical approaches

Abstract This study investigated the groundwater chemistry, suitability, and potential human health risk in the southwest part of Bangladesh. Groundwater samples were collected from the shallow aquifer throughout the study area. A set of different hydro-chemical analyses, water quality indices, multivariate statistics, and geo-statistical models were applied to achieve the study objectives. Study results show the concentration of NH3-N, HCO3 −, Ca2 +, As, Fe, and Mn exceeded Bangladesh drinking water standards in 94%, 100%, 100%, 73%, 97%, and 91% of samples, respectively. Groundwater quality indices indicate that about 94% of samples are suitable for irrigation, and about 82% are unsuitable for drinking. In the study area, groundwater is mainly Ca2+-Mg2-HCO3 − types, and rock-water interactions dominate the mineralization process. Ca2+  > Mg2+  > Na+  > K+, and HCO3 −  > Cl−  > SO4 2− are the sequential orders of major cation and major anion, respectively. Multivariate analyses show the geogenic origin is predominant over anthropogenic sources. Semivariogram models show moderate to weak spatial dependence. The non-carcinogenic and carcinogenic health risks via oral exposure pathways for adults and children are in the high (97%) and very high (100%) categories, respectively.


Introduction
Safe and sustainable quality water is vital for life's existence, ecological stability, and economic progress worldwide. Groundwater is considered a more suitable option than surface water for drinking and irrigation purposes in many countries of the world due to its prompt availability, good access, and free from contamination (Bhuiyan et al. 2016, Şener et al. 2017, Batabyal 2018). Groundwater quality is significantly influenced by natural processes (e.g. breakdown of patent material, rock water interaction, and leaching/ infiltration from the soil surface) and anthropogenic activities (e.g. agricultural activities, landfills, municipal wastes, and rapid urbanization and industrialization) (Bhuiyan et al. 2016, Wagh et al. 2017. The changes in groundwater quality and quantity also depend on hydrogeological factors and seasonal variation (Thivya et al. 2014, Rahman andRahaman 2018), as well as the movement and bioavailability of chemical elements in groundwater being controlled by pH, alkalinity, oxidation, and reduction potential, dissolved organic matter, and ion exchange capacity (Markich et al. 2001, Musa et al. 2013. Hydro-chemical studies of groundwater offer comprehensive knowledge regarding the potential changes in quality, hydrochemical processes controlling parameters (Varol and Davraz 2015). Therefore, periodic monitoring of the water quality parameters is needed for sustainable groundwater use in diverse fields, including drinking, domestic, agriculture, and industrial activities (Şener et al. 2017, Ahmed et al. 2019, Bodrud-Doza et al. 2019). Usually, groundwater has different ions, metals, and metalloids, which have significant health problems if exposed to an excessive amount for a long time (Varol and Davraz 2016). In the world, approximately 80% of diseases and one-third of deaths in developing countries are caused by polluted drinking water (WHO 2004). Several studies reveal that concentrations of As, Fe, and Mn were higher than other trace metals in the groundwater of Bangladesh (Islam et al. 2015, Ahmed et al. 2019, Ghosh et al. 2020, Chakraborty et al. 2022. Due to exposure to the elevated levels of As, Fe, and Mn in drinking water for an extended period, different types of health hazards might be created, such as neurological, cardiovascular, hematological, renal, respiratory, neurological disorders, etc. (Islam et al. 2017a(Islam et al. , 2017bGhosh et al. 2020). Since contaminated drinking water is directly connected to human health, a methodology-based risk assessment approach is applied to identify and characterize the qualitative (chronic, carcinogenic) and quantitative (calculation and mapping) risk levels (S¸ener et al. 2017). On the other hand, Bangladesh is an agricultural country where irrigation water quality is vital for agricultural activities. Poor quality irrigated water causes enormous economic loss by reducing crop production and deteriorating soil fertility. So, long-term agricultural productivity significantly depends on suitable irrigation water (Islam et al. 2017a(Islam et al. ,2017b). In Bangladesh, groundwater is directly used for drinking (about 95% of rural and 70% of urban people) and irrigation (80%) purposes (WHO and UNICEF 2017, Islam et al. 2018a, 2018b), but for several reasons such as overuse, changing of river flow direction, inadequate management of water resources, climatic factors, and anthropogenic activities make this groundwater resource more vulnerable (Bodrud-Doza et al. 2016, Ahmed et al. 2020. Much consideration has been needed to monitor groundwater quality because it influences human health and social-economic developments. Traditional methods for assessing water quality, such as trilinear plots and statistical approaches, are widely used (Kumar et al. 2015). However, many researchers have recently focused on determining groundwater quality, quantity, hydrogeochemical heterogeneity, pollution source identification, modeling of groundwater movement, solute transport, and leaching by employing various types of water quality indices, geographic information system (GIS), and multivariate statistical methods (Islam et al. 2017a, 2017b, Şener et al. 2017, Islam et al. 2018a, 2018b, Honarbakhsh et al. 2019. Therefore, water quality indices, along with GIS and multivariate statistical approaches, are one of the most effective tools for water quality management among environmental managers, decision-makers, and water planners (Shabbir and Ahmad 2015).
This study was carried out in Jhikargachha Upazila, Jashore district, situated in the southwest part of Bangladesh. It is an essential province of the Jashore district, especially in flower cultivation and other agricultural activities. The study area uses groundwater as drinking, cooking, domestic, and irrigation water. In this circumstance, it is essential to know the present status of the groundwater in the study area. These findings will help determine whether it requires further treatment or not for using this groundwater for drinking and irrigation purposes. So, the objectives of this study are (i) to investigate the hydro-chemical characteristics of groundwater; (ii) to determine the groundwater suitability for drinking and irrigation purposes; and (iii) to assess the potential human health risks caused by exposure to As, Fe, and Mn through groundwater ingestion.

Study area
The Jhikargachha Upazila, under Jashore District, is located in the southwestern part of Bangladesh and lies between 89 00 0 to 89 07 0 E longitude and 22 55'55" to 230 12'34" N latitude ( Figure 1). The geology of this area is characterized by Pleistocene-Modhupur clay, and the thicknesses of underlying fluvial-deltaic sediments are Holocene age that formed the main aquifer system (Ahmed et al. 2020). The hydrogeological state of the study area is categorized into three layers (Jakariya et al. 2003). The three lithological indices are designated as topsoil, clay, or silty clay, to silt and sand. The lithological distribution is also indicated in such a way that clay-fine sandmedium sand-porous medium sand (Ahmed et al. 2020). Jhikargachha is rarely flooded and is geo-morphologically more stable than the area mainly underlain by the active Meghna floodplain. The whole of Jhikargachha Upazila is underlain by Holocene-Recent fluvial (river) sediments (GSB 1990). Jhikargachha is likely underlain by the full range of fluvial sediments (gravels, sands, silts, and clays) related to different relict features of fluvial systems such as in-filled oxbow lakes, floodplains, meander belts, and levees (Jakariya 2000). The annual average temperature ranged from 15.4 to 34.6 C (59.7-94.3 F), and the annual rainfall totaled 1537 mL (60.5 inches) (Jakariya 2000). However, the total area of the Upazila is 308.09 km 2 , with a population of 271,014, and the main water bodies are the Kobadak and Betna River, while this Upazila consists of eleven unions. Agriculture (65.97%) is the primary source of income in the study area. The sources of drinking water for this study area are tubewell (96.18%), tap water (0.57%), pond water (0.18%), and others (3.07%) (Banglapedia 2014).

Sample collection, preparation, analysis, and quality assurance
In this study, 33 groundwater well samples were collected from the Jhikargachha Upazila of Jashore district in Bangladesh (Figure 1), from April to May 2019. In the study area, tube wells were installed between 10 and 15 years, ranging from 2001 to 2015, with the median year being 2004. The depths of the sample collected tube wells ranged from 100 to 220 ft. (30.48-67.05 m), with the mean value being 154.8 ft (47.18 m), which the tube-well owners recorded. Groundwater samples were collected in 500-ml polystyrene bottles. Sampling bottles were washed with 1:1 HNO 3 and rinsed three times with distilled water. Before sampling, the tube wells were pumped for 15-20 min, and then water samples were collected in a pre-cleaned sampling bottle after being rinsed with sample water three times. These water samples were subjected to both field (pH, electrical conductivity [EC], and total dissolved solids [TDS]) and laboratory analyses (PO 4 3À , NO 3 À , SO 4 2À , NH 3 -N, HCO 3 À , Cl À and hardness as CaCO 3 , As, Fe, Mn, Na þ , K þ , Mg 2þ , and Ca 2þ ). The field analyses were measured with calibrated portable instruments. The water sample pH was measured by a pH meter (Model: MARTINI instruments, pH 56 pHWP, USA), EC and TDS were measured using a Conductivity Meter (HACH Sension À156; multi-parameter, USA). The pH meter was initially calibrated with three buffer solutions at pH 4.0, pH 7.0, and pH 10.0, and then the pH meter was verified after measuring three samples. For EC and TDS determination, the multi-meter was calibrated using 1000 lS/ cm EC and 1000 mg/L TDS standard solutions and verified after three measurements. To preserve the samples for cations and trace metal analysis, one bottle was acidified with concentrated HNO 3 (69%, Merck, Germany), and the other was kept un-acidified for anions and other analyses. All water samples were labeled, kept in a cooler box, shifted to our laboratory, and then kept in a freezer at 4 C until further chemical analysis was performed. Sample collection procedures and preservation techniques are adopted from Huang et al. (2018). HCO 3 À , Cl À , and hardness as SO 4 2À (SulfaVer V R 4 Reagent Powder Pillows; Method: 8051; 680 Program; detection ranges: 2-70 mg/L) and NH 3 -N (Nessler Method: 8038; 380 Program; detection ranges: 0.02-2.50 mg/L) in groundwater samples were measured by spectrophotometer (HACH DR 39000, USA). The concentration of As, Fe, Mn, Na þ , K þ , Mg 2þ , and Ca 2þ in the samples were analyzed by using a Hydride Generator, and an air-acetylene flame Atomicabsorption-spectrophotometer (AAS) system (Model: AA-7000, SHIMADZU, Japan), equipped with a single element hollow-cathode lamp as a light source at the wavelength of 193.70, 248.30, 279.5,589.00, 766.50, 285.20, and 422.70 nm, respectively. The concentration limit of detection for As was 0.0003 mg/L, while for Fe, Mn, Na þ , K þ , Mg 2þ , and Ca 2þ was 0.01-0.004 mg/L and the quality assurance of these parameters is presented in Table S1. The instrument calibration was done by using different concentrations of working solution from the standard solution (1000 ppm), obtained from Sigma Aldrich, Switzerland. The analysis results of all parameters were expressed as mg/L without EC (lS/cm). Deionized ultrapure water was utilized for the whole experimental work. All laboratory equipment and glassware were cleaned (i.e. used 20% HNO 3 acid for cleaning purposes and frequently washed with double distilled water and oven-dried) before use. Every sample was made to run duplicate analyses, including blank, and validated for quality assurance. Based on the necessity, water samples were diluted several times, and the relative standard deviations of detected major ions and elements were within ± 5-7%.

Drinking water suitability
The weighted arithmetic water quality index (WAWQI) method (Yisa and Jimoh 2010) is used to evaluate drinking water suitability. This index is calculated using Equations (1-3), respectively.
where S i , C i , and Q i specify that ith parameter standard value, ith parameter experimental concentration, and scale of quality rating, respectively. Equation (2) measures relative weight: where the i parameter's standard value is inversely proportional to the relative weight. Finally, the total WQI according to Equation (3) was calculated: The detailed procedures of the WAWQI method have been explained in Rahman and Rahaman (2018). Relative weight and categories of water quality are presented in Tables S2 and 2 respectively.

Irrigation water suitability
The irrigation water quality index (IWQI) is an integrated approach used to assess groundwater's suitability for irrigation purposes. This method is adopted from Ayers and Westcot (1985), and detailed explanations are presented by Islam et al. (2017aIslam et al. ( , 2017b. The calculation methods are given in Equations (4-5).
The irrigation quality index is calculated by Equation (4): where i is an incremental index and G represents the contribution of each one of the four hazard categories that are important to assessing the quality of an irrigation water resource. G can be calculated by using Equation (5): where k is an incremental index, N is the total number of parameters available for the analysis, w is the weight value of the groups, and r is the rating value of each parameter.

Irrigation water suitability evaluation indices
The concentration of EC is applied to assess irrigation suitability concerning the salinity of groundwater. Furthermore, several indices are used to determine irrigation water quality, including total hardness (TH), sodium adsorption ratio (SAR), Kelley's ratio (KR), sodium percentage (%Na), permeability index (PI), magnesium adsorption ratio (MAR), and residual sodium bicarbonate (RSBC). All ionic concentrations are in milli equivalents per liter (meq/L) except TH (mg/L). The categories of these indices are presented in Table 4.
The irrigation water quality indices are calculated using Equations (6-12). However, the TH (Todd 1980) is calculated by Equation (6): According to US Salinity Laboratory (Richards 1954), SAR is expressed as Equation (7): Lastly, KR (Kelley 1963) as described as Equation (8): The Na % is calculated as Equation (9) (Todd 1980): Doneen (1964) defined PI as described in Equation (10): MAR (Ragunath 1987), also known as magnesium hazard (MH) was calculated as Equation (11): The RSBC is expressed as Equation (12) (Gupta 1983): All these parameters, as well as individual chemical parameters, have been compared with national and international standards to assess the groundwater for suitability of irrigation water.

Health risk assessment
This study applied the United States Environmental Protection Agency (USEPA) proposed non-carcinogenic risk (NCR) and the cancer risk (CR) methods for assessing children and adults health risk through ingestion of drinking water.
CDI is the chronic daily intake (mg/kg/d) (US Environmental Protection Agency [USEPA] (1989) and calculated by using Equation (13).
where C represents the heavy metal concentration (mg/L), IR the drinking water ingestion rate in L/d (3.53 L/d for adults (Milton et al. 2006), and 1.0 L/d for children (USEPA 1989); ED the exposure duration in years (70 years for adults and 6 years for children) (USEPA 1989); EF the exposure frequency in days/year (365 d for adults and children) (USEPA 1989); BW the average body weight (50 kg for adults and 15 kg for children) (USEPA 1989) , and AT is the averaging time HQ is used to assess the NCR associated with noncarcinogenic contaminant exposure (USEPA 2001, USEPA 2004, Islam et al. 2017a, 2017b. HQ is calculated by using Equation (14).
The NCR is evaluated by the Hazard index (HI) presented in Equation (15).
The CR is calculated by multiplying the chronic daily intake (CDI) by the slope factor (SF) of cancercausing heavy metals, as shown in Equation (16).
where SF is the slope factor of contaminants (mg/Kg/ d) (1.5 mg/Kg/d for As) (USEPA 1989). The classification of NCR and CR is detailed in Bodrud-Doza et al. (2019).

Statistical and geostatistical analysis
In this study, multivariate analysis tools such as Pearson's correlation matrix (PCM), principal component analysis (PCA), cluster analysis (CA), hydro-geochemical characterizations, and physicochemical parameters data are analyzed by SPSS version 0.20 (SPSS Inc., Chicago, IL) and Microsoft Excel 2010. The ordinary kriging (OK) and semivariogram models have been applied to the spatial distribution of the groundwater quality evaluating indices and human health risk, performed by ArcGIS version 10.5). The literature illustrates the interpolation methods (Masoud 2014, Tapoglou et al. 2014).

Characterization and drinking water suitability of groundwater
To assess the groundwater suitability for drinking purposes, the analyzed physicochemical parameters are compared with the WHO (2011) and BDWS (1997) guidelines. The pH values varied from 6.8 to 7.9, with an average value of 7.13 ± 0.26, representing the slightly alkaline nature of groundwater (Table 1). EC is a significant parameter for drinking water quality, indicating the dissolved solids' levels and the source water's ionic strength (Ahmed et al. 2019). The concentration of EC in the water sampled varied from 350.00 to 2090.00 ms/cm with an average value of 633.93 ± 327.41 ms/cm. On the other hand, TDS in the study area water samples was 405.72 ± 209.54 mg/L, ranging from 224.00 to 1337.6 mg/L (Table 1). According to EC and TDS classification, almost all groundwater samples are in the freshwater category (Freeze and Cherry 1979). The concentrations of dissolved cations are Ca 2þ , Mg 2þ , Na þ , and K þ ranged from 80.21to 209.31, 19.11 to 39.16, 9.98  acceptable limits, but NH 3 -N exceeds the recommended level due to degradation of naturally occurring organic matter or anthropogenic sources. The concentration of heavy metals and metalloids in groundwater samples followed the order Fe > As > Mn, respectively (Table 1). The measured concentrations of As ranged from 0.013 to 0.501 mg/L with a mean value of 0.09 ± 0.08 mg/L (Table 1). According to the study results, about 97% and 73% of groundwater samples exceeded the WHO (0.01 mg/L) and BDWS (0.05 mg/L), respectively. The mean concentration of As was 9 and 1.8 times higher than WHO (0.01 mg/L) and BDWS (0.05 mg/L), respectively. Iron concentrations in groundwater water showed a wide range (0.30-7.81 mg/L) with an average concentration of 3.68 ± 1.72 mg/L (Table 1). Almost 97% of the groundwater samples exceeded the BDWS and WHO permissible limits. The average value of iron in the study area groundwater was about 12 and 3.68 times higher than the WHO (0.3 mg/L) and the higher limit of BDWS (0.3-1.0 mg/L), respectively. The Mn concentration in groundwater samples varied from 0.02 to 0.63 mg/L with an average concentration of 0.38 ± 0.19 mg/L (Table 1). This study revealed that approximately 45% and 91% of groundwater samples exceeded the WHO (0.4 mg/L) and BDWS (0.1 mg/L), respectively. The mean value of Mn was 3.8 times higher than the BDWS (0.1 mg/L) and 1.3 times higher than the former WHO health guideline (0.4 mg/L). Ghosh et al. (2020) and Jakariya (2000) found high concentrations of As, Fe, and Mn in the study area. The WAWQI is further used to measure the overall quality of water for drinking purposes. In this study, all analyzed physicochemical parameters are utilized for calculating the WAWQI at each sampling point.
The WAWQI values ranged from 33.64 to 846.60, with a mean value of 187.60 ± 134.21 ( Table 2). The finding shows that only 3% of the sampling area is in the good category, while 82% is in the unsuitable category. Hence, the groundwater of these areas is not suitable for drinking purposes. Usually, skewness should be within the standard range ± 2; then, it is measured as an extreme (Reimann et al. 2008). This study shows EC, TDS, Na þ , Cl À , NO 3 À and As in the positively skewed datasets and are considered extreme for groundwater aquifer systems. Similarly, for kurtosis, these parameters, namely EC, TDS, Naþ, Cl À , NO3 À , and As, are leptokurtic (value > 3), whereas the rest of the parameters concentrations are platykurtic (Table 1).

Hydro-geochemical facies and water type
The hydro-chemical facies is a vital feature for determining groundwater hydrochemistry. The Piper diagram is an effective and widely used method for characterizing groundwater hydrochemistry using cation and anion concentrations. Figure 2(a) shows that Ca 2þ , Mg 2þ , and HCO 3 À are essential in describing the study area's groundwater quality. Based on the analyzed water sample results, this study area groundwater is Ca 2þ -Mg 2þ -HCO 3 À type, which means that Ca 2þ , Mg 2þ ,and HCO 3 À are dominating elements compared to Na þ , K þ , Cl À , and SO 4 2À . The Ca 2þ -Mg 2þ -HCO 3 À water type is related to the area where the base rock-groundwater inter-relation is the main influencing factor for the difference in the groundwater chemistry from the hydrologic basins (Yidana 2010). However, the Ca 2þ -Mg 2þ -HCO 3 À facies is derived from the dissolution of calcite, which exists on the limestone of the Eocene age in the aquifer (Rahman et al. 2017). A Gibbs plot is applied to assess the prime processes that control the hydro-geochemistry. Figure 2(b) shows that all groundwater samples fall in the rock dominance region, which is controlled by the processes of carbonate mineral dissolution within the aquifer. Dissolution of rock is an important process contributing to understanding groundwater hydrochemistry. The rock dominance in the groundwater chemistry is also confirmed by the computation of Hounslow ratios (Hounslow 1995): when the value of Cl -/ P all anion ¼ 0.11 (Table 1), less than 0.8  indicates mainly weathering of rock, especially carbonate rock (Rahman et al. 2017). The results are consistent with the findings for groundwater resources in the northwestern part of Bangladesh (Bhuiyan et al. 2015) and northern Bangladesh (Howladar et al. 2017).

Human health risk assessment
The higher concentration of heavy metals (such as As, Fe, and Mn) in drinking water can create different human health problems. In this study, the non-carcinogenic and carcinogenic risks to children and adults are estimated due to ingestion of heavy metals (e.g. As, Fe, and Mn) via drinking water. The assessment results for children and adults are presented in Table  3. The HQ value of Fe and Mn in water samples for children and adults is less than the recommended level (HQ > 1), demonstrating that Fe and Mn have no significant non-carcinogenic health risks, but the HQ value of As exceeded the recommended level (HQ > 1), suggesting that As may pose a non-carcinogenic health risk to children and adults. The HI value shows the integrated effect of multiple contaminants. In this study, HI values varied from 3.03 to 112.25 with a mean value of 20.54 ± 18.41 and from 3.21 to118.87 with a mean value of 21.75 ± 19.50 for children and adults, respectively, indicating that both children and adults have high non-carcinogenic health risks such as vomiting, abdominal pain, and diarrhea; injury to the skin, gastrointestinal, and respiratory tract, damage to the liver, cardiovascular, hematopoietic, and nervous system; diabetes; hair loss; and neurological problems, etc. The mean CR values for children and adults are 9.01 and 9.54E-03, indicating that of every 10 000 people in the study area, 9.01 and 9.54 people would have CR (Table 3). It has been observed that the descending order of CR among the three groups was adults > children. The CR values of both groups (children and adults) exceed the USEPA (2001) recommended safe ranges (1E-06 to 1E-04), suggesting that each group (children and adults) has the possibility of carcinogenicity. However, this study did not mention any particular types of arsenic-associated cancer disease in the study area. Several authors mentioned that skin, kidney, and bladder cancers are the general types of cancer due to chronic exposure to arsenic-contaminated drinking water (Sadeghi et al. 2018). Smith et al. (2000) found that consumption of arsenic in water was about 500 mg/L; by age 60, more than 1 in 10 had developed skin cancer. In the Jashore district, about 51% of surveyed tube-well water contains more than 50 mg/L of arsenic (Smith et al. 2000). About 50 million people in Bangladesh were expected to be at risk of exposure to arsenic through the consumption of water from contaminated tube wells. The common indicators of arsenicosis reported in Bangladesh include melanosis (98.9-100.0%), keratosis (58.8-92.7%), leucomelanosis (29.2-42.7%), weakness (88.2-93.0%), chronic cough (20.0-33.8%), conjunctival congestion (9.4-25.0%), and non-pitting edema (2.8-6.9%) (Ahmad et al. 1997).

Irrigation water quality indices
The IWQI is applied to evaluate the suitability of groundwater for irrigation purposes in the study area. The IWQI calculated value ranged from 23 to 29.5, where about 31 sampling points (94%) are in highly suitable categories (IWQI > 23), indicating that this study area's groundwater is highly suitable for irrigation purposes (Table 2). EC is vital for categorizing irrigation water; meanwhile, it measures the TDS as well as the salinity of groundwater. Based on EC standards for irrigation water quality (Table 4), most of the sample's (79%) EC values are in the range of 250-750 lS/ cm, suggesting good quality for irrigation. The calculated TH indicated that all the samples were in the very hard quality category (Table 4). The idea of SAR is applied to measure the possible sodium hazard because it assesses the soil sodium adsorption capacity from irrigated water. High SAR values terminate the soil structure by influencing the cation exchange process in the soil. The SAR values ranged from 0.19 to 1.89 meq/L with a mean value of 0.49 ± 0.44, indicating good quality irrigation water for this study area (Tables 1 and 4). KR shows groundwater samples' relationship among Na þ , Ca 2þ , and Mg 2þ ions. KR > 1 indicates not suitable for irrigation due to the present excess level of Na þ in water, and KR < 1, indicates suitable for irrigation. The attained KR values varied between 0.03 and 0.52 meq/L ( Table 1), suggesting that all sampling stations are in a "suitable" class (Table 4). Na% indicates the amount of sodium in irrigation water. The higher amount of sodium in the irrigation water hampered the plant's growth rate. The Na% value ranges between 4.25 and 35.39 (Table 1). About 91% of the water sample is in the "excellent" category, and 9% is in the "good" category (Table 4). PI helps to determine the soil quality, which may be affected by long-term irrigation. PI is categorized into three classes, including Class 1 (PI > 75%), Class 2 (25% < PI < 75%), and Class 3 (PI < 25%). According to Table 4, all samples fall into the class 1 category, suggesting that the research area is suitable for longterm irrigation. MAR shows a negative impact on the soil if it surpasses 50. In the study area, MAR varied from 16.42 to 39.07, with an average value of 26.69 ± 6.02, representing no negative impacts on the soil for irrigation (Table 4). Totally, 100% of samples have less than 50 values, implying that these study areas' groundwater has no risk of infiltration from other aquifers' soil (Table 4). The RSBC is classified into three groups < 5, 5-10, and < 10 meq/L, measured as safe, marginal, and unsatisfactory. The RSBC values ranged from À5.52 to 4.50 meq/L with a mean value of À0.72 meq/L (Table 1), representing that sample water is appropriate for irrigation. The results of Mg 2þ /Ca 2þ and Na þ /Ca þ ratios varied from 0.11 to 0.38 with a mean of 0.22 ± 0.06 and 0.06 to 0.97 with a mean value of 0.20 ± 0.22, respectively (Table 1), indicating no threat of infiltration problems for soil from the groundwater in the study area.

Source of ions and factors controlling groundwater quality
Multivariate analyses (PCA, PCM, and CA) are used to determine groundwater quality parameters' source, distribution, and geochemical interaction (Tables S3  and S4; Figure S1, respectively). This study successively used PCA, PCM, and CA to detect the probable sources of groundwater quality parameters. For PCA, seven factors with eigenvalues greater than one are drawn out for groundwater data sets, which show 82.15% of the total variance. In this study, PC1, PC2, PC3, PC4, PC5, PC6, and PC7 illuminate more than 25.61%, 15.81%, 11.97%, 8.49%, 7.15%, 6.97%, and 6.13% of the total variance, respectively (Table S1). The PC1 is loaded with EC, TDS, and Cl -, which are also located in the same cluster ( Figure S1(a)) and represent a significant positive correlation among them (r 2 [EC-TDS ¼ 1.00] and [EC-Cl -¼0.92]; Table S4). The highest value of EC confirms the geogenic process which leads to the storage of salts in soils rather than anthropogenic activities such as agricultural practices and fertilizer uses (Drever 1997, Jiang et al. 2009). These salts reach the groundwater through the infiltrated recharge water. The presence of Na þ and K þ in the same cluster reveals their common existence and origin, which can be attributed to weathering of Na þ and K þ bearing rock minerals or leaching of Na and K-rich fertilizer from the soil horizon to the aquifer (Subba et al. 2014). This affects the water quality and might be regulated by ionic or reverse ionic exchange in the study area (Loni et al. 2015 Table S4) and the similar cluster position ( Figure S1(a)). The wastes and agrochemicals are the probable sources of NO 3 À and PO 4 3À in the groundwater. The high pH value could be due to carbonate rock dissolution in groundwater, where alkalinity conditions control the groundwater aquifer system. The high HCO 3 À the value may be contributed by long-term irrigation practices that circulate the water within the soil/weathered zone. PC4 is positively loaded with As and Fe (r 2 [As-Fe ¼0.50], Table S4), along with the same cluster (Figure Table 4. Irrigation water quality indices for the study area. S1(a)), which may be generated from the rock-water interaction (e.g. weathering of arsenopyrite, ferruginous quartzite, pyrites) and leaching of secondary salt through rainwater. Arsenic in groundwater release from the Holocene alluvial/deltaic sediments is due to the strongly reducing nature of groundwater in Bangladesh (Smedley andKinniburgh 2002, Jakariya et al. 2003). The PC5 only loading with SO 4 2À is suggestively influenced by anthropogenic inputs such as agrochemicals (Todd 1980). PC6 is dominated by NH 3 -N, Mn, and Mg 2þ . The location of NH 3 -N and Mn in the same cluster ( Figure S1(a)) shows their common existence and origin, while Mg 2þ originated from different sources. The possible sources of Mg 2þ in the groundwater are dissociating dolomite and limestone rock (Lasaga 1984). Positive loading of Mg 2þ indicates ionic exchange geochemistry. On the other hand, NH 3 -N and Mn probably come from the degradation of organic matter and rock-water interaction (Bodrud-Doza et al. 2016). Finally, PC7 is loaded by hardness with a separate cluster that may be generated from weathering of sedimentary rock and calcium-bearing minerals or extreme uses of lime in agricultural activities. Despite some variations, the multivariate analysis represents good agreement for identifying groundwater quality parameter sources. Finally, PCA results showed that geogenic sources (rock weathering and ion exchange), followed by anthropogenic activities (domestic sewage and agro-chemicals), and were the main controlling factors for the groundwater quality of the study area. Additionally, the results of PCA are validated using the CA and PCM analyses, which also match with previous findings of Bhuiyan et al. (2016). Q mode CA is applied to identify the spatial likenesses and sampling site groupings in the study area. This study represents two clusters, where cluster one consists of eight sampling sites (S1-3, S26, S28, S30, S31, and S33) at a low link distance, which is connected to cluster two, twenty-five sampling sites (S4-25, S27, S29, and S3) at a high link distance ( Figure S1(b)), indicating that the majority of samples follow the high link distance due to different parameters.

Geostatistical model
The nugget/sill (N/S) ratio designates the spatial dependence of groundwater quality indices and human health risk. Three categories (N/S ratio >25%, 25-75%, and <75% indicate strong, moderate, and weak spatial dependence, respectively) are applied to illustrate the models (Shi et al. 2007). The experimental, omnidirectional, average of the semivariogram shows a binned sign, blue line, and plus sign in the semivariogram model, respectively (Figure 3). The circular semivariogram model is identified to be the best-fitted model for EC, SAR, and Na (%) values. The exponential semivariogram model fitted best for IWQI, and TH values. The Gaussian semivariogram model fitted best for WAWQI, HI children, HI adults, CR children, and CR adults, respectively (Table S5). The ordinary kriging results exhibited that EC, WAWQI, HI children, HI adults, CR children, and CR adults values have weak spatial dependence (Figure 3(a, c, g-j)), while IWQI, SAR, Na (%), and TH show a moderate spatial dependency (Figure 3(b, d-f)). The semivariogram models show moderate to weak spatial dependence, which may be due to the variation of topography, soil characteristics, localized effects of anthropogenic activity, land use patterns, prevailing aquifer geology, and long-term geogenic processes such as groundwater source rock, rainfall, infiltration processes, etc.

Spatial distribution maps
The spatial distribution of WAWQI, IWQI, EC, SAR, Na (%), and TH is shown in Figure 4, as well as the noncarcinogenic and carcinogenic health risks of children and adults in Figure 5. Based on the IWQI map, $94% of groundwater in the study area is suitable for irrigational uses (Figure 4(b)), and the map shows the central part is a more distributed area. On the other hand, EC confirmed an increasing trend from the northern to the southern direction (Figure 4(c)) due to significant variation of Na þ and Cl À in the studied area. The spatial map of SAR and Na (%) showed higher values in the central and eastern parts, but the reverse in the northern and southern parts of the study area (Figure 4(d,e)). The higher TH values in the northwest part of the study area (Figure 4(f)) are probably due to the weathering of sedimentary or sedimentary rock and leaching of lime from the agricultural land soil surface to the groundwater aquifer. The spatial maps of WAWQI displayed that about 82% of the area is not suitable for drinking purposes where the WAWQI values increased in the northwest to the southern region of the study area (Figure 4(a)), which may be due to the highly influenced geogenic and manmade activities in the aquifer system. Such an outcome is found in the similar work of Islam et al. (2015). Generally, the spatial distribution of HI and CR values for children and adults is highly distributed throughout the study area. However, the study area's southwest part is more critical (Figure 5).

Conclusions
This study assessed the groundwater suitability and human health risk in the southwest part of Bangladesh. Groundwater in shallow aquifers is slightly alkaline with a TDS > 600 mg/L. The sequence of major cation and anion in the groundwater samples are Ca 2þ > Mg 2þ > Na þ > K þ , and HCO 3 À > Cl À > SO 4 2À , respectively. The hydro-geochemical facies of groundwater is dominantly of the HCO 3 À Ca 2þ ÁMg 2þ type. Multivariate analysis (PCA, CA, and PCM) shows that geogenic origin is highly responsible for the variation of studied parameters in the study area. Semivariogram models show moderate to weak spatial dependence. According to IWQI, about 94% of the samples went to high suitability water types. Several irrigation quality indices (SAR, Na%, PI, MAR, RSBC, KR, and TH) indicate that studied groundwater is suitable for long-term irrigation with no salinity and sodium hazards. Regarding drinking water suitability, about 82% of the sampling area's groundwater is not suitable for drinking. The inhabitants of the study area (children and adults) have higher potential non-carcinogenic and carcinogenic health risks due to the consumption of heavy metals (As, Fe, and Mn) via drinking water. Special consideration must be given to arsenic because its hazard quotient is very high, making the groundwater unsuitable for direct ingestion without further treatment. To improve this situation, this study recommended numerous initiatives such as sealing the unsafe tube wells, installing deep tube wells, implementing low-cost water treatment facilities (e.g. community-based or household level) for contaminated tube well water; allowing venders to supply safe water at a low price.