Groundwater samples were collected from ten major towns (eleven samples because 2 from Mahendranagar) in the central to far west Terai region of Nepal during May 2025 (Figure 1). We decided to take one sample from each town of this region and from Mahendranagar we took two samples such as Mahendranagar-B representing an urban center and Mahendranagar-G representing an agricultural field to see urban influence or agricultural influence for comparison. The exact construction date was not available but based on migration data those hand pumps might be installed sometime between three to four decades before. People use water for domestic use, including drinking without any treatment. More detailed sample description with some physical parameters is given in Table 1. Physical parameters, including water temperature, pH, electrical conductivity (EC), and total dissolved solids (TDS), were measured on-site using a Hanna Instruments handheld multimeter.

Table 1. Description of sampling sites with physical parameters of groundwater along central to far west region of Terai in Nepal.

For major solute and nutrient analyses, samples were collected in 125 mL acid-washed polyethylene bottles, filtered through 0.7 μm precombusted (5+ hours at 450˚C) Whatman GF/F glass microfiber filters, and refrigerated locally before being shipped on ice to the New Hampshire Water Resources Research Center at the University of New Hampshire (Durham, NH, USA). This suite of analyses included major base cations (Na⁺, K⁺, Mg²⁺, Ca²⁺), ammonium ( NH 4 + ), major anions (Cl^-, NO 3 − , SO 4 2 − , HCO 3 − , PO 4 3 − ), dissolved organic carbon (DOC), total dissolved nitrogen (TDN), sodium adsorption ratio (SAR), and boron. For trace element analysis, separate groundwater samples were filtered through 0.45 μm polycarbonate membrane filters, acidified with nitric acid and shipped to Carroll College (Helena, MT, USA) to quantify arsenic (As), cadmium (Cd), cobalt (Co), chromium (Cr), copper (Cu), iron (Fe), nickel (Ni), lead (Pb), antimony (Sb), selenium (Se), and zinc (Zn).

Water samples were analyzed for major cations, major anions, nutrients, dissolved organic carbon (DOC), total dissolved nitrogen (TDN), and trace metals. Major cations (Na⁺, K⁺, Mg²⁺ and Ca²⁺), and major anions (Cl^-, Br^-, NO 3 − , and SO 4 2 − ), were measured via ion chromatography with suppressed conductivity using a Metrohm 940 Professional IC Vario. Cation analysis utilized an oxalic acid mobile phase in accordance with ASTM (2007) [38] guidelines, while anion analysis followed US EPA (2007) Method 300.1 [39]. Nutrients were analyzed by automated colorimetry using a SmartChem discrete analyzer; ammonium was determined via the automated phenate hypochlorite method with sodium nitroprusside enhancement (US EPA 2005, Method 350.1) [40], and orthophosphate ( PO 4 3 − ) was determined via the automated ascorbic acid reduction method (US EPA 2005, Method 365.1) [41]. Dissolved organic carbon was analyzed as non-purgeable organic carbon using a Shimadzu TOC-L analyzer following US EPA (2002) Method 415.1 [42], where samples were injected into a 720˚C furnace with a platinum catalyst on silica beads to convert carbon compounds into CO₂ for quantification via a non-dispersive infrared detector (NDIR). Total dissolved nitrogen was quantified using a Shimadzu TOC-L system following the method of Merriamet al. (1996) [43], which converts nitrogen compounds to NO at 680˚C for detection by a chemiluminescent nitrogen detector. Finally, dissolved organic nitrogen (DON) was calculated by subtracting inorganic nitrogen ( NO 3 − -N + NH 4 + -N ) from the total dissolved nitrogen concentration.

Water samples were analyzed at the New Hampshire Water Resources Research Center following strict quality assurance and quality control (QA/QC) protocols for major ions, nutrients, and dissolved carbon. To ensure data reliability, instrument calibration models were constructed using 4-to-7-point curves, and Method Detection Limits (MDLs) were defined at the 99% confidence threshold. This yielded target analyte detection limits of 0.08 mg/L for Cl⁻, 0.004 mg/L for NO 3 − , 0.06 mg/L for SO 4 2 − , 3.4 µg/L for PO 4 3 − , 5 µg/L for NH 4 + , 0.03 mg/L for Na⁺, 0.003 mg/L for K⁺, 0.05 mg/L for Mg²⁺, 0.16 mg/L for Ca²⁺, 0.05 mg/L for DOC, and 0.20 mg/L for TDN. Analytical accuracy was systematically monitored via certified Quality Control Samples (QCS; Ultra Scientific or SPEX Certiprep) analyzed at a frequency of approximately every 10 - 15 samples within each analytical batch. Across all measured parameters, the laboratory enforced a uniform acceptable precision limit of ±15% for laboratory duplicates and an accuracy recovery limit of ±15% for quality control standards.

Trace elements (As, Cd, Co, Cr, Cu, Fe, Ni, Pb, Sb, Se, and Zn) were analyzed using an inductively coupled plasma-optical emission spectrometer (ICP-OES; PerkinElmer Avio 200) according to standard methods for determining metals in environmental samples (US EPA 1991) [44]. The instrument was operated under standard conditions utilizing argon plasma, auxiliary, and nebulizer gas flows, with measurements collected in axial view. Quantification was achieved via external calibration curves prepared from multi-element ICP standards, all of which exhibited excellent linearity (R² ≥ 0.995). To ensure analytical accuracy and instrument stability, procedural blanks and calibration verification standards were analyzed periodically. The instrument detection limits (mg/L) for the target elements were 0.025 for As, 0.001 for Cd, 0.002 for Co, 0.001 for Cr, 0.009 for Cu, 0.001 for Fe, 0.002 for Ni, 0.002 for Pb, 0.015 for Sb, 0.030 for Se, and 0.002 for Zn.

To provide an integrated assessment of drinking water suitability across sampling sites, a Weighted Arithmetic Water Quality Index (WQI) was calculated using selected physicochemical parameters and trace elements that have established drinking-water guideline values. The WQI converts multiple water quality measurements into a single dimensionless score that facilitates comparison among sites and provides an overall indication of drinking water quality. Parameters were selected based on their relevance to drinking-water safety and availability of guideline values from the World Health Organization (WHO 2022) [45] and Nepal Drinking Water Quality Standards (NDWQS 2025) [46]. Variables without established health-based drinking standards were excluded to avoid disproportionate influence and preserve interpretability of the index.

Relative weights were assigned to reflect the relative importance of each parameter to drinking-water quality and potential human health impacts. Higher weights were assigned to parameters with stronger health implications, particularly toxic trace elements. The WQI was calculated using the weighted arithmetic approach according to the following equations.

First, the relative weight for each parameter was computed as:

where RW_i is the relative weight assigned to parameter i, w_i is the assigned importance weight, and n is the total number of included variables.

The quality rating for each parameter was then calculated as:

where Q_iis the quality rating, C_i is the measured concentration of parameter i, and S_i is the corresponding drinking-water standard.

For pH, quality rating was calculated relative to the ideal value of 7.0:

The sub-index for each parameter was estimated as:

Finally, the overall water quality index for each site was computed as:

Lower WQI values indicate better drinking water quality.

Table 2 summarizes the physicochemical characteristics, nutrient concentrations,

Table 2. Descriptive statistics and bootstrap 95% confidence intervals for physicochemical parameters, nutrients, major ions, and trace elements in drinking water samples from the Central-Far Western Terai region of Nepal.

NB: Units for all elements are reported in mg/L, except EC (μS/cm); pH and SAR are unitless. NG = No health-based guideline established; NS = Not specified in NDWQS; NA = Not available. NDWQS = Nepal Drinking Water Quality Standards. WHO = World Health Organization.

major ions, and selected trace elements measured in eleven drinking water samples collected during May 2025 from the Central-Far Western Terai region of Nepal. For each parameter, the table reports the sample size (n), minimum and maximum values, arithmetic mean, standard deviation (SD), bootstrap-based 95% confidence interval (CI) for the mean, and the corresponding drinking-water guideline values established by the World Health Organization (WHO, 2022) [45] and the Nepal Drinking Water Quality Standards (NDWQS, 2025) [46].

The sampled waters exhibited near-neutral pH, ranging from 6.6 to 7.2 with a mean of 6.90 ± 0.18, well within the acceptable limits established by both WHO and NDWQS (6.5 - 8.5). The narrow bootstrap confidence interval (6.78 - 7.02) indicates little variation in pH among the sampling sites. Electrical conductivity (EC) ranged from 655 to 1596 μS/cm (mean = 972 μS/cm), whereas total dissolved solids (TDS) varied from 327 to 814 mg/L with a mean of 507 mg/L. The mean TDS was well below the NDWQS guideline value of 1000 mg/L, indicating acceptable levels of dissolved minerals. However, EC exceeded the NDWQS guideline (1500 μS/cm) at two locations (Lamahi and Kohalpur), suggesting localized increases in dissolved ionic constituents that warrant continued monitoring.

Nitrogen species showed greater spatial variability than the general physicochemical parameters. Dissolved organic carbon (DOC) remained low (mean = 0.42 mg/L), whereas total dissolved nitrogen (TDN) displayed the greatest variability (mean = 2.78 ± 3.70 mg/L), reflecting elevated concentrations at only a few sampling sites. Total inorganic nitrogen (TIN) constituted the major fraction of dissolved nitrogen (mean = 2.19 mg/L), while dissolved organic nitrogen (DON) contributed a smaller proportion (mean = 0.58 mg/L). The relatively wide bootstrap confidence intervals for TDN and TIN further indicate spatial heterogeneity in nitrogen concentrations across the study area.

The major anions occurred at concentrations well below drinking-water guideline values. Chloride ranged from 1.72 to 210.09 mg/L (mean = 35.89 mg/L), remaining below the WHO and NDWQS limit of 250 mg/L. Sulfate concentrations were also low (mean = 5.74 mg/L), and nitrate averaged only 2.11 mg/L, far below the drinking-water standard of 50 mg/L. Phosphate and bromide concentrations were similarly low throughout the study area, indicating no evidence of widespread contamination from major inorganic anions during the sampling period.

Among the major cations, sodium showed the greatest variation (3.74 - 72.67 mg/L; mean = 25.42 ± 20.83 mg/L), although all observed concentrations remained well below the WHO advisory value of 200 mg/L. Potassium concentrations were consistently low (mean = 2.44 mg/L), whereas magnesium exhibited considerable spatial variation (mean = 19.32 ± 15.78 mg/L), likely reflecting differences in local geology and groundwater-rock interactions. Calcium averaged 14.08 mg/L, substantially below the NDWQS guideline value of 200 mg/L.

Ammonium concentrations were uniformly low (mean = 0.08 mg/L), well below the NDWQS guideline value of 1.5 mg/L. The sodium adsorption ratio (SAR) ranged from 0.132 to 1.350, with a mean of 0.78, indicating a very low sodium hazard. Although SAR is commonly used to assess irrigation suitability rather than drinking water quality, these values suggest that the sampled waters would also be suitable for agricultural use.

The selected trace elements (Fe, Zn, Cu, and Ni) were detected at low concentrations and complied with the applicable drinking-water guideline values. Iron averaged 0.032 mg/L, approximately one-tenth of the WHO and NDWQS guideline value (0.30 mg/L). Zinc, copper, and nickel also remained well below their respective guideline values in all samples, indicating no evidence of trace-metal contamination in the analyzed drinking water.

Because only eleven water samples were available, conventional large-sample confidence interval methods may produce unstable estimates and rely on distributional assumptions that are difficult to verify. To obtain more reliable interval estimates, bootstrap resampling with 10,000 replications was used to construct 95% confidence intervals for the samples mean. Bootstrap methods estimate the sampling distribution empirically through repeated resampling with replacement and are well suited for small environmental datasets. Using 10,000 replications reduces Monte Carlo error and provides stable confidence interval estimates without requiring strict normality assumptions.

Because only eleven water samples were available and several variables were not normally distributed, relationships among water-quality variables were evaluated using Spearman’s rank correlation coefficient (ρ), a non-parametric measure of monotonic association that is less sensitive to small sample size and distributional assumptions than Pearson’s correlation. To improve clarity, only statistically significant correlations (p < 0.05) are presented. Correlations among mathematically derived nitrogen variables (TDN, TIN, DON, and NO₃) are omitted from interpretation because these variables are not statistically independent and their strong associations largely reflect their computational relationships rather than independent hydrochemical processes.

The correlation analysis identified several statistically significant associations that are consistent with the hydrochemical characteristics of the sampled groundwater (Table 3). The strongest relationship was observed between electrical conductivity (EC) and total dissolved solids (TDS) (ρ = 0.964, p < 0.001), confirming that dissolved ionic constituents were the primary contributors to water conductivity which is obvious. Electrical conductivity was also positively associated with magnesium (ρ = 0.618), indicating that dissolved alkaline-earth ions contributed substantially to the overall mineralization of the groundwater.

Strong positive correlations were also observed among several major ions. Chloride exhibited a strong association with bromide (ρ = 0.856), while sodium was highly correlated with magnesium (ρ = 0.864) and the sodium adsorption ratio (SAR) (ρ = 0.936). These relationships suggest common geochemical controls,

Table 3. Statistically significant Spearman rank correlations among selected drinking water quality variables in the Central-Far Western Terai region of Nepal.

including mineral dissolution and groundwater-rock interaction, which influence the ionic composition of the water.

Among variables, dissolved organic carbon (DOC) showed positive correlations with chloride (ρ = 0.609), bromide (ρ = 0.687), ammonium (ρ = 0.624), and SAR (ρ = 0.764). In contrast, DOC exhibited negative correlations with arsenic (ρ = −0.627) and antimony (ρ = −0.691). These associations may reflect differences in geochemical conditions controlling the occurrence and mobility of organic matter and trace elements within the groundwater system.

Sulfate was positively associated with calcium (ρ = 0.636), supporting the influence of mineral weathering, but negatively correlated with phosphate (ρ = −0.622), cobalt (ρ = −0.630), and selenium (ρ = −0.724). In addition, pH showed significant negative correlations with chromium (ρ = −0.715) and nickel (ρ = −0.686), suggesting reduced mobility of these metals under less acidic conditions. Several significant relationships were also identified among trace elements. Iron and nickel were also strongly negatively correlated (ρ = −0.765), whereas cadmium and chromium exhibited a positive association (ρ = 0.695).

Correlations involving mathematically derived nitrogen variables (TDN, TIN, DON, and NO 3 − ) were not interpreted because these variables are not statistically independent. Their strong correlations primarily reflect their computational relationships and serve as a check on the internal consistency of the analytical measurements rather than providing evidence of independent hydrochemical processes.

To provide an integrated evaluation of drinking water quality, a Weighted Arithmetic Water Quality Index (WQI) was calculated for each sampling site using physicochemical parameters and trace elements with established drinking-water guideline values. Only parameters with recognized standards from the World Health Organization (WHO, 2022) [45] or the Nepal Drinking Water Quality Standards (NDWQS, 2025) [46] were included in the index calculation. Following data verification, only reliably quantified parameters were retained, whereas variables without established drinking-water standards or excluded during quality assurance were omitted. The WQI was computed using the relative parameter weights, quality ratings, and sub-index values described in Sub-Section 3.3. Parameters with greater public health significance, such as nitrate and nickel, were assigned higher weights than general indicator parameters (Table 4(a)). Lower WQI values indicate better overall drinking water quality.

Table 4(b) summarizes the WQI values for the eleven sampling sites together with the overall summary statistics. WQI values ranged from 7.28 to 26.12, with a mean of 14.02 ± 5.43 and a coefficient of variation of 38.7%. According to the adopted classification scheme, all sampling sites were classified as Excellent (WQI < 50), indicating favorable chemical water quality based on the physicochemical parameters and selected trace elements included in the WQI calculation. Because WQI is based only on the measured chemical parameters, it should not be interpreted as a comprehensive assessment of drinking-water safety, particularly since microbiological quality and other potential contaminants were not evaluated spatially and seasonally from the region.

Among the sampling locations, Mahendranagar-G recorded the lowest WQI (7.28), representing the best overall water quality, whereas Lamahi had the highest WQI (26.12). Although some spatial variation was observed among sampling sites, even the highest WQI remained well below the threshold for the Excellent category. The WQI results are consistent with the descriptive statistics presented in Table 2. The relatively higher WQI values observed at a few locations were primarily associated with elevated electrical conductivity, total dissolved solids,

Table 4. (a). Parameters, assigned weights, relative weights, drinking-water standards, reference sources, and justification used for Water Quality Index (WQI) calculation; (b) Water Quality Index (WQI) values, drinking water quality classification, and summary statistics for sampling sites in the Central-Far Western Terai region of Nepal.

Note: Water Quality Index (WQI) classification: <50 = Excellent; 50 - 100 = Good; 100 - 200 = Poor; 200 - 300 = Very Poor; >300 = Unsuitable for drinking.

and major ion concentrations rather than exceedances of health-based drinking-water standards. Overall, the WQI assessment confirms that the drinking water sources sampled in the Central-Far Western Terai region during May 2025 were of excellent quality and suitable for human consumption based on measured chemical parameters.

To complement the statistical analyses presented above, boxplots were prepared for six representative drinking water quality indicators: electrical conductivity (EC), nitrate ( NO 3 − ), sodium (Na), magnesium (Mg), nickel (Ni), and the Water Quality Index (WQI). These variables were selected because they represent overall mineralization, nutrient status, major ion chemistry, a regulated trace element of public health relevance, and the overall quality of drinking water. The boxplots provide a visual comparison of the distributions, variability, and potential outlying observations among the eleven sampling sites.

Figure 2 illustrates the variability of the selected water quality indicators across the study area. The WQI exhibited the smallest spread, with all values well below the threshold for the Excellent category (WQI < 50). This observation is consistent with the results presented in Table 4(a) and Table 4(b) and confirms that drinking water quality was uniformly good across all sampling locations based on the measured chemical parameters. Nickel concentrations showed only a narrow range of variation among the samples and remained well below the WHO guideline value of 0.020 mg/L. The absence of unusually high values indicates that nickel contamination was not a concern in the sampled drinking water sources during the study period.

Nitrate concentrations displayed moderate variability, with most samples containing relatively low concentrations and a few sites exhibiting slightly higher values. However, all nitrate concentrations remained substantially below the WHO and NDWQS drinking water standard of 50 mg/L, indicating no evidence of widespread nitrate contamination. Among the major physicochemical indicators, EC, sodium, and magnesium exhibited greater variability than the other measured parameters, reflecting differences in dissolved mineral content among sampling locations or effect of human activities primarily from hotel waste as most of the sites are from town center. Despite this variation, the observed concentrations generally remained within the applicable drinking water guideline values.

Overall, the boxplots complement the descriptive statistics by showing that, although some spatial variation exists among the sampling sites, the measured water quality parameters remained within acceptable limits. The consistently low WQI values further support the conclusion that the drinking water sources sampled during May 2025 were of excellent quality and suitable for human consumption based on the measured chemical parameters.

Figure 2. Boxplots of selected drinking water quality parameters across sampling sites in the Central-Far Western Terai region of Nepal.