Assessing the water quality of River Ganga in Varanasi, India, through WQI, NPI, and multivariate techniques: a comprehensive study Open Access

Indian rivers are facing a threat because the discharge of various contaminants through various industries, agricultural runoff, untreated sewage, and improper disposal of solid waste has led to deterioration in the quality of river water (Gupta et al. 2020a,2020b; Ali et al. 2021). River Ganga at Varanasi is equally threatened by these pollution sources such as partially treated sewage, cremation of dead bodies, and various religious and commercial activities (Girija et al. 2007; Narain 2014; Marathe et al. 2017). Despite the high levels of contamination in the river water, it is still used for various purposes, which could potentially harm human health. Therefore, determining the quality of the water is crucial in raising public awareness. Water-borne diseases affect 80% of Indians, and deaths from diarrhoea and other water-borne diseases are increasing (Conaway 2015).

Reports from the Central Pollution Control Board of India showed that almost half of the sewage generated in cities along the banks of the Ganga is discharged into the river without proper treatment, leading to adverse impacts on the river (CPCB 2021). Varanasi, one of the major cities situated along the River Ganga, has been reported to have the worst water quality among all the cities situated along the Ganga. The city generates an estimated 230–410 million litres per day of municipal sewage, which is discharged into the River Ganga after the treatment, along with other sources such as industrial effluents, household waste, and remains from cremated bodies (Chaudhary et al. 2017). These discharges have a negative impact on the physicochemical and biological qualities of the river, and the problem is expected to worsen in the future (Sarkar et al. 2012; Chaudhary et al. 2017).

Despite having some information about the water quality of the Ganga River in Varanasi, there is still a lack of comprehensive and long-term studies. Therefore, there is a pressing need for a detailed evaluation of the water quality of the Ganga River at Varanasi (Das & Tamminga 2012). Measuring water quality by quantifying numerous physicochemical parameters is a highly challenging task, which nevertheless only provides a loose approximation of the overall situation. Therefore, an effective evaluation tool is needed that can integrate all the relevant parameters to represent the water quality as a single numerical value (Gupta et al. 2019). Among various evaluation tools, the water quality index (WQI) is the most commonly used tool for water quality assessment. It considers all the relevant parameters, provides appropriate importance to each parameter, and compares them to the standards set by government authorities to protect public health (Ewaid et al. 2020; Nong et al. 2020). WQI is a comprehensive scientific technique for determining the overall state of water quality to determine appropriate treatment strategies to address current concerns (Tyagi et al. 2013; Jain 2020). WQI is one of the most recent approaches to study the water quality as a whole (Kamboj & Kamboj 2019; Dimri et al. 2020; Gupta et al. 2020a, 2020b). The Nemerow pollution index (NPI) serves as a crucial tool in evaluating and quantifying environmental pollution levels, offering a comprehensive assessment of the impact of various pollutants on ecosystems. Developed by Leonard Nemerow, this index amalgamates multiple parameters into a single metric, to gauge the extent of pollution, encompassing diverse factors such as air, water, and soil quality. Its versatility and ability to encapsulate complex environmental data make the NPI a valuable instrument in environmental research, guiding efforts aimed at mitigating pollution and preserving ecological balance.

Multivariate methods such as cluster analysis (CA) and principal component analysis (PCA) are utilized in different studies to recognize potential pollutants (Misaghi et al. 2017). PCA is a multivariate technique, and in situations where colossal measure of information is accessible, it is a reasonable way to reduce the data (Gupta et al. 2020b). Moreover, CA is additionally one of the multivariate techniques used to evaluate relative similarity in the homogeneity of estimated parameters (Shrestha & Kazama 2007). Multiple researches on the water quality of the Ganga and its tributaries have been published using WQI and multivariate statistical methods (Kamboj & Kamboj 2019; Dimri et al. 2020; Jain 2020; Matta et al. 2022; Nandi et al. 2022). The earlier published work on water quality assessment of headwater stream (Bhagirathi) and the Ganga in the Himalayan region has been reported only on some and limited study sites (Sarin et al. 1992; Singh & Hasnain 1998; Chakrapani 2005; Sood et al. 2008; Chakrapani & Saini 2009; Matta et al. 2015), which does not indicate a pattern of impact on water quality at a specific location within the entire stretch from Varanasi upstream to downstream. The city of Varanasi, nestled along the banks of the sacred Ganges River, grapples with the intricate challenges of water quality. Despite its spiritual significance and cultural prominence, Varanasi faces acute issues regarding the Ganga's water pollution. The river here encounters a myriad of pollutants stemming from urban waste, industrial discharge, untreated sewage, and religious rituals performed directly at its ghats (Dimri et al. 2020). Consequently, the water quality of the River Ganga at Varanasi is a significant concern, affecting the health of both the river ecosystem and the millions who rely on its waters for various purposes. Efforts to address this issue have been initiated, encompassing local initiatives, government interventions, and public awareness campaigns. However, the persistence of water quality challenges in Varanasi underscores the urgency for sustained, comprehensive strategies to restore and maintain the Ganges's health in this historically and spiritually significant city.

The goals of the present study were to evaluate the water quality of the Ganga River in Varanasi, using physicochemical analysis, WQI, NPI, and multivariate statistics, and identify the most significant factors, i.e., (i) identifying association or disparity among study sites and different seasons, (ii) identifying water quality parameter variations in river water quality, and (iii) identifying the impact of possible pollution sources (natural or anthropogenic).

The current investigation was carried out for a period of 2 years in the monsoon, pre-, and post-monsoon seasons from nine different locations from the entry point to downstream up to the limits of Varanasi city, stretched across a length of around 12 km. All the samples were collected at a fixed date every month between 08.00 AM and 10.00 AM. Samples were collected, preserved, transported, and analysed per standard methods prescribed in APHA (2017). The samples were analysed for a total of 20 different (physicochemical, biological) parameters. Unstable factors such as pH, temperature, total dissolved solids (TDS), and dissolved oxygen (DO) were evaluated by using a portable multiparameter water quality analysis kit (Hanna HI98194) immediately after sampling. The results of the qualitatively analysed samples are presented in Table S1.

All of the apparatus used for the experiments had been pre-calibrated using a standard solution or in accordance with the standard guideline to guarantee the accuracy and precision of the observations. Standard operating procedures for all the equipment and chemical analyses were adhered to throughout the study period with the necessary safety precautions. All the chemical analyses were carried out using ‘A’ grade glassware with chemicals from Borosil, Merck, and Fisher Scientific to ensure accurate results.

The descriptive statistics (minima, maxima, mean, and standard deviation) were computed for all the parameters using Excel 2019. A violin plot was created with box plots of the biological data for each site using Origin Pro 2023. The data distribution was examined by using the Kolmogorov–Smirnov test. A correlation matrix plot of the variables was produced using the Spearman correlation test because most of the data did not follow a normal distribution. Correlation matrix plots, PCA, and heat maps with dendrogram analysis were carried out using R 4.2.2.

To evaluate the potential impact of different parameters on the hydrochemistry of the river and to identify implicit linear connections from a variety of variables, PCA was used (Simeonov et al. 2003). The correlation matrix between a parameter and the data matrix was used to standardize the data matrix prior to statistical analysis to make certain that all parameters were handled similarly and that no parameter with higher actual values dominates the PCA (Simeonov et al. 2004). The principal components were created in a sequential order, with decreasing contributions to variance, for example, the first principal component (PC1) significantly contributes to variations in the initial results, and subsequent component (PC2) accounts for decreasing proportions of variance (Vieira et al. 2012).

CA (also known as clustering) has been used to arrange the large datasets into clusters according to a predetermined set of shared criteria. The fundamental goal of clustering was to categorize groups or clusters of monitoring stations that were comparably similar based on their similarities and differences. In this study, the CA based on the hierarchical approach for standardized datasets was used. A visual representation of the grouping process, the dendrogram, shows the clusters and their surroundings while significantly reducing the complexity of the original data (Simeonov et al. 2003).

Spearman's correlations were utilized to investigate the correlation structures between the variables that showed non-normal distribution of the parameters affecting water quality (Shrestha & Kazama 2007). By examining Spearman's correlation coefficient (Spearman's R), these non-parametric correlations might be used to further examine the temporal fluctuations of the river water quality metrics (Varol & Sen 2009; Wang et al. 2013).

A violin plot is a type of data visualization that shows the distribution of data across different categories. It is similar to a box plot in a way that it not only displays the median, quartiles, and range of the data but also shows the probability density of the data at different values, providing a more complete picture of the data's distribution. The shape of the violin plot is generally symmetrical around the median, with wide sections indicating high density of the data and narrow sections indicating lower density. The plot can be used to compare the distribution of data across multiple groups or categories and can provide insights into the skewness, kurtosis, and other characteristics of the data.

The weighted arithmetic WQI (WAWQI), which gives each criterion a weight that represents how essential it is in determining the quality of the water, was first proposed by Horton in 1965. This WQI was developed using the Indian criteria and World Health Organization (WHO) standards (WQI).

Subsequently, a quality rating (q_i) was calculated for each parameter using the following equation:

• w_i α 1/S_i

• w_i=k/S_i

• k = Proportionality constant and w_i = Unit weightage

• k = 1/∑ (1/S_i).

The order of the water quality parameters measured in this study, based on the impact on human health, is likely to be as follows: ⁠. For the WQI computation, weight was determined based on the standard value of each parameter. The guidelines of BIS IS 10500:2012 and water quality criteria B from CPCB were followed in the current study's standard values for each parameter (http://www.uppcb.com/river quality.htm). The relative parameter weights are presented in Table 1, and the categorization of WQI for drinking water quality suitability is presented in Table 2.

Table 1

Relative parameter weights

Sl. No. .	Parameters .	Standard value (Si) (BIS/CPCB) .	Unit weight (Wi) .
1	pH	8.5	0.17
2	EC	1,000	0.01
3	TDS	500	0.002
4	TH	300	0.004
5	TA	200	0.007
6	DO	5	0.29
7	BOD	3	0.48
8	COD	250	0.005
9	Cl−	250	0.005
10		50	0.02
11		200	0.007
12	TC	500	0.002

Sl. No. .	Parameters .	Standard value (Si) (BIS/CPCB) .	Unit weight (Wi) .
1	pH	8.5	0.17
2	EC	1,000	0.01
3	TDS	500	0.002
4	TH	300	0.004
5	TA	200	0.007
6	DO	5	0.29
7	BOD	3	0.48
8	COD	250	0.005
9	Cl−	250	0.005
10		50	0.02
11		200	0.007
12	TC	500	0.002

View Large

Average temperatures of the pre-monsoon, monsoon, and post-monsoon river water were 22.61 ± 0.90, 24.69 ± 1.03, and 19.92 ± 0.82 °C, respectively. A change in temperature with the season has been encountered, which shows the impact of seasonality on water temperature. The depth, turbulence, time of day, and heat input from the outside environment were all observed to influence water temperature. Temperature plays a key role in various natural processes of the aquatic environment. A change in the temperature may trigger a change in various parameters.

The average pH value was observed as 8.11 ± 0.16, 7.87 ± 0.13, and 8.17 ± 0.11 during the pre-monsoon, monsoon, and post-monsoon seasons, respectively, indicating that the water was slightly alkaline (Figure 2). During the pre-monsoon and monsoon season, the maximum pH was as 8.23 (S1) and the minimum was 7.90 (S9). The same order was found in the monsoon and post-monsoon seasons, i.e., the maximum value was 8.07 at S1 and the minimum was 7.68 at S9. In the post-monsoon season, the maximum value was 8.25 (S1) and the minimum value was 7.92 (S9). According to BIS (2012) guidelines, the standard range of the pH value is 6.5–8.5, which was followed in the samples. However, the water was found to be alkaline in nature. The influence of domestic and industrial effluents entering the rivers and the activities of photosynthetic algae, which consume dissolved CO₂, are the causes of the alkaline nature of the river water.

Electrical conductivity (EC) varies considerably in different sites. The EC of the water samples on different sites was recorded as 537.17 ± 43.66, 411.69 ± 33.45, and 495.74 ± 62.09 μS/cm during the pre-monsoon, monsoon, and post-monsoon seasons, respectively (Figure 2), and comparatively higher values were obtained in the pre-monsoon season at S9 because of the wastewater from Varuna Nala and some domestic sewage from Shahi Nala. During the pre-monsoon season, EC was 597.67 μS/cm (S9) and the minimum value was 492 μs/cm (S6). In the post-monsoon season, the maximum value was 602.60 μS/cm (S9) and the minimum value was 441.63 μS/cm (S5). Higher values of EC were found in the post-monsoon season followed by the pre-monsoon and monsoon seasons owing to the increase of ionic concentration because of the religious human activities, industrial wastes, and draining of contaminated water through drains and other sources into the River Ganga (Dimri et al. 2020).

The average value of the TDS under the present study was observed as 270.37 ± 21.97, 208.25 ± 15.63, and 248.92 ± 29.75 mg/L for the pre-monsoon, monsoon, and post-monsoon seasons (Figure 2). The highest value of TDS during the pre-monsoon season was 300.50 mg/L (S9) followed by 294.75 (S2), 293.0 (S8), and 278.75 mg/L (S1), respectively. During the post-monsoon season, the maximum TDS was 236.88 mg/L (S8) while the minimum value was 186.81 mg/L (S1). Higher TDS values were observed at locations S9, S8, and S2 because of the discharge of sewage and cremation remains from the surroundings. During the monsoon season, TDS remained in the favourable condition because of water dilution. The higher value of TDS in the post-monsoon season than in pre-monsoon and monsoon seasons of the Ganga River can be attributed to a combination of factors including reduced flow rate, evaporation, resuspension of bottom sediments, groundwater inputs, and anthropogenic activities (agricultural runoff, industrial effluents, and untreated sewage).

Salinity values were 0.24 ± 0.01, 0.21 ± 0.02, and 0.32 ± 0.11 PSU during pre-monsoon, monsoon, and post-monsoon seasons, respectively. Saltiness depends upon the path flow and the contacting environment of the water along with the solubility of minerals from aquifers and surrounding rocks. Soluble salts in the surface water originate primarily from the solution of rock materials and domestic waste. Hence, salinity usually increases with the increasing depth. In the pre-monsoon season, the maximum salinity was 0.32 PSU (S9) and the minimum value was 0.20 PSU (S1), while in the monsoon, the maximum salinity was 0.26 PSU (S9) and the minimum value was 0.19 PSU (S3). During the post-monsoon, the maximum salinity was 0.42 PSU (S9) and the minimum value was 0.21 PSU (S1). The combination of reduced freshwater inflow, increased evaporation, tidal influence, anthropogenic activities, and groundwater inputs leads to higher salinity in the River Ganga during the post-monsoon season compared with the pre-monsoon and monsoon seasons.

The average alkalinity values were 167.14 ± 3.13, 124.35 ± 3.98, 166.91 ± 6.29 mg/L during the pre-monsoon, monsoon, and post-monsoon seasons, respectively. In the pre-monsoon season, the maximum alkalinity was 174.12 mg/L (S2) and the minimum value was 164.58 mg/L (S7), while in the monsoon season, the maximum alkalinity value was 129.88 mg/L (S2) and the minimum value was 116.54 mg/L (S4). In the post-monsoon season, the maximum alkalinity was 173.12 mg/L (S9) and the minimum value was 159.70 mg/L (S3). Bicarbonate was the main species causing alkalinity in about 80–90% of the sample because of the bicarbonate dissolved in the water, which reacts with the water to form carbonic acid (H₂CO₃), which then dissociates to form a bicarbonate ion. This is likely due to the fact that bicarbonate is a very common component of natural waters and can be present in significant concentrations. In addition, bicarbonate is relatively stable in water and does not undergo rapid changes in concentration or reactivity. Overall, the presence of bicarbonate and its contribution to alkalinity can have important implications for water quality and treatment. For example, alkalinity can affect the pH of water and the effectiveness of certain disinfection methods. According to BIS (2012), the limit of alkalinity in consumable water is 200 mg/L. Under the present study, all the values were under permissible limits. The average values of hardness were observed as 199.30 ± 4.07, 146.76 ± 2.99, and 207.28 ± 8.28 mg/L in the pre-monsoon, monsoon, and post-monsoon seasons, respectively. During the pre-monsoon season, the maximum value was 202.55 mg/L (S9) and the minimum value was 195.98 mg/L (S4), while in the monsoon period, the maximum value was 152.74 mg/L (S9) and the minimum value was 144.59 mg/L (S6). In the post-monsoon season, the maximum value was 211.88 mg/L (S9) and the minimum value was 202.58 mg/L (S4). The main contributors to water hardness are anions such as carbonates, bicarbonates, and chloride, as well as cations such as calcium and magnesium. Numerous human health issues, including kidney, heart, and stone problems, are brought on by the hardness of water (Kamboj & Kamboj 2019). According to BIS (2012), the permissible level of hardness in drinking water is 300 mg/L. Under this study, all the values were under permissible limits. The flowrate of the River Ganga decreases significantly, leading to a higher concentration of dissolved ions. This concentration increase is particularly pronounced for ions that contribute to alkalinity and hardness, such as calcium, magnesium, and bicarbonates. The extensive contact between water and rocks/sediments increases the dissolution of minerals that contribute to alkalinity and hardness, such as limestone and dolomite. Evaporation can further concentrate dissolved ions in the river water, leading to higher alkalinity and hardness values. The combined effects of these factors result in higher alkalinity and hardness values in the River Ganga during the post-monsoon season compared with that during the pre-monsoon and monsoon seasons (Kumar et al. 2015; Dimri et al. 2020; Trombadore et al. 2020; Nandi et al. 2022).

A high value of chloride in water is an indicator of the water pollution contributed by organic waste with either industrial or animal origins. The average values of chloride were observed as 55.66 ± 4.95, 37.51 ± 7.40, and 51.22 ± 5.46 mg/L in the pre-monsoon, monsoon, and post-monsoon seasons, respectively. During the pre-monsoon season, the maximum chloride was 58.28 mg/L (S8) and the minimum level was 45.43 mg/L (S5), while in the monsoon period, the maximum level was 42.89 mg/L (S9) and the minimum level was 25.64 mg/L (S7). In the post-monsoon season, the maximum level was 59.37 mg/L (S9) and the minimum chloride value was 45.07 mg/L (S1). The permissible limit of chloride in consumable water is 250 mg/L (BIS 2012). All the values for chloride observed in the present study were under the permissible limit (BIS 2012). Pre-monsoon has a higher chloride concentration than post-monsoon, and this might be due to the high temperature, reduced flow, and reduced discharges into the river. A high value of chloride was observed at sites S8, S9, and S2, which might be because of the discharge of household waste, sewage, and industrial discharge (Kumar et al. 2006).

Dissolved oxygen is a key characteristic in determining the quality of the water because it reflects the underlying physical as well as biological processes in water bodies. Under the present investigation, the average DO values were 7.18 ± 0.41, 7.70 ± 0.29, and 7.27 ± 0.50 mg/L in the pre-monsoon, monsoon, and post-monsoon seasons, respectively. Due to the increased oxygen solubility in water at lower temperatures, DO levels were high in the monsoon season. In addition, greater photosynthetic activity brought on by ample sunlight and the visibility of clear water is what causes oxygen to be released into water (Ravindra & Kaushik 2003). The lower value of DO in the pre-monsoon season may be because of the high rate of evaporation and sedimentation processes and high turbidity, leading to lesser saturation of light in the water.

The biochemical oxygen demand (BOD) values observed in the water samples collected throughout different seasons were found to be higher than the permissible limit of 3 mg/L (BIS 2012), with the pre-monsoon, monsoon, and post-monsoon average values being 3.59 ± 0.53, 2.93 ± 0.47, and 3.69 ± 0.45 mg/L, respectively. In the present study, the highest BOD value was observed at S9 (Varuna Confluence Zone) in all the three seasons owing to the direct discharge of sewage, open drainage household discharge into the river from Varuna, Shahi Nala, and agricultural runoff. The findings of the study revealed that the BOD values were comparatively higher than those reported in a few earlier studies (Gupta et al. 2017).

Chemical oxygen demand (COD) gives an overall assessment of the organic matter present in water. COD is very useful for determining the contamination level of sewage and industrial waste. The average COD values were 32.61 ± 2.53, 26.51 ± 2.94, and 33.85 ± 2.62 mg/L during the pre-monsoon, monsoon, and post-monsoon seasons, respectively. The highest value of COD in the pre-monsoon season was 38.78 mg/L (S9) and the minimum level was 29.97 mg/L (S5), while in the monsoon season, the maximum level was 28.03 mg/L (S2) and the minimum level was 23.43 mg/L (S7). During the post-monsoon season, the maximum level was 40.38 mg/L (S9) and the minimum level was 31.47 mg/L (S1). The higher concentration was observed in the post-monsoon than in pre-monsoon and monsoon seasons. The COD values were found to be higher in downstream sites, possibly because of partially treated sewage and industrial effluents. The previous reports recorded COD value of 27.81 mg/L in the Ganga River at Varanasi (Kumari et al. 2013). Our reported values are slightly higher than some of the previously reported values for River Ganga at Varanasi. The higher values of BOD and COD in the post-monsoon season compared with the pre-monsoon and monsoon seasons of Ganga water can be attributed to several factors such as anthropogenic activities (untreated waste discharge, agricultural runoff, urbanization, and industrial and hospital wastewater), reduced flow rate, increased organic matter input, and reduced self-purification capacity. Average nitrate concentrations in the present study were 1.46 ± 0.65, 0.89 ± 0.29, and 1.17 ± 0.05 mg/L in the pre-monsoon, monsoon, and post-monsoon seasons, respectively. The end products of the metabolic oxidation of ammonia are represented by the geographical and temporal fluctuation in nitrates (Mahananda et al. 2010).

The average phosphate concentrations were 0.53 ± 0.48, 0.19 ± 0.14, and 0.38 ± 0.24 mg/L during the pre-monsoon, monsoon, and post-monsoon seasons, respectively. During the pre-monsoon season, the maximum level was 1.60 mg/L (S2) and the minimum level was 0.17 mg/L (S1 and S6), while in the monsoon period, the maximum level was 0.48 mg/L (S2) and the minimum level was 0.04 mg/L (S7). In the post-monsoon period, the maximum concentration was 0.93 mg/L (S1). In soluble and organic forms, phosphate can be found in natural water systems (Hasan et al. 2009). The use of detergents is known to notably escalate the phosphate content in water. The quality of river water is significantly impacted by the addition of man-made phosphorus to surface water, primarily originating from domestic sewage. The variation in phosphate levels across seasons in the river could be influenced by several factors, including agricultural runoff, sediment release, and varying human activities, contributing to fluctuations in phosphate concentrations observed during different seasons.

The average concentrations of sulphate were 30.01 ± 1.81, 25.59 ± 1.96, and 35.06 ± 1.59 mg/L in the pre-monsoon, monsoon, and post-monsoon seasons, respectively. During the pre-monsoon period, the maximum level was 31.88 mg/L (S8) and the minimum level was 26.31 mg/L (S3), while in the monsoon season, the maximum level was 27.54 mg/L (S6) and the minimum level was 22.05 mg/L (S3). In the post-monsoon period, the maximum level was 38.08 mg/L (S9) and the minimum level was 33.52 mg/L (S3). The permissible limit of sulphate in water is 200 mg/L (BIS 2012). All the values for sulphate in the present study were under the permissible limit (BIS 2012). The lower value of sulphate observed may be due to its tendency to easily precipitate and settle to the bottom of the river (Abdul-Razak et al. 2010).

The 0–0.5 (0 to −0.5) correlation coefficient shows negligible correlation to moderate positive (negative) correlation, and the 0.5–1 (−0.5 to −1) correlation coefficient shows moderately positive (negative) to very high positive (negative) correlation (Mukaka 2012). The temperature was negative correlated with total hardness (−0.64) and weak negative correlated with DO (−0.32). The pH moderately correlated with DO (0.57) and alkalinity (0.52), and weakly negatively correlated with temperature (−0.31). EC correlated very highly correlated with TDS (0.99) and moderately with alkalinity (0.69), chloride (0.67), and sulphate (0.55). TDS indicated a moderate correlation with alkalinity (0.68), chloride (0.66), and sulphate (0.66). Alkalinity had a strong positive correlation with hardness (0.79) and a moderately positive correlation with chloride (0.56) and faecal coliform (0.53), respectively. Total hardness moderately correlated with sulphate (0.52). Chloride correlated moderately with phosphate (0.55), weakly correlated with sulphate (0.31), and negatively correlated with nitrate (−0.07). DO correlated negatively moderately with BOD (−0.69), and it also negatively correlated with FC (−0.46) and TC (−0.46). There was a significant high positive correlation observed between BOD and both FC (0.74) and TC (0.74). FC correlated very highly with TC (0.92) (Figure 4).

Dendrograms were also constructed to explain the relationship between all the data points in the system. Under this study, the dendrogram represented three groups of monitoring sites, and in each group, all sampling locations followed this trend without any exception. In the pre-monsoon period (Figure 5(a)), cluster I is generated by sites S1, S3, S4, S5, S6, and S7, cluster II by sites S8 and S9, and cluster III by S2. In the monsoon period (Figure 5(b)), cluster I is generated by S1, S3, S4, and S5, cluster II by sites S6, S7, and S8, and cluster III by S2 and S9. In the post-monsoon (Figure 5(c)) season, cluster I is generated by S1, S3, S4, S5, S6, and S7, cluster II by sites S8 and S9, and cluster III by S2. These clusters received open drainage waste, domestic and hospital sewage, cremation remains, automobile and small industrial waste, and agricultural runoff (Sharma et al. 2021). Dendrograms of the pre- and post-monsoon having almost the same trends compared with the monsoon season.

PCA is a mathematical approach that does not rely on assumptions and is used to minimize the number of samples while maintaining the original sample information (Gupta et al. 2020b).

Each NPI value, less than or equal to 1, represents the relative pollution that each parameter has produced.

The order of the water quality parameters measured in this study, by their impact on human health, is likely to be as follows: ⁠. For the WQI computation, weight was determined based on the standard value of each parameter. The guidelines of BIS IS 10500:2012 and water quality criteria B from CPCB were followed in the current study's standard values for each parameter (http://www.uppcb.com/riverquality.htm). Different sources of pollution in the study sites are direct discharge from open drains, untreated sewage, industrial influent, hospital and domestic sewage, agricultural runoff, and cremation remains.

The findings of the present study revealed that the concentration of physicochemical parameters in the water of the River Ganga was higher downstream than it was upstream, indicating that River Ganga at Varanasi is negatively impacted by activities such as discharge of domestic sewage, open drain, and agricultural and industrial effluents because of extensive urbanization. Various sources have resulted in higher coliform, high pH, low DO, high BOD, and COD. The observed WQI values subsequently indicated that the water quality of the Ganga was degraded to the level that it was unsafe for bathing. The WQI values played a crucial role in identifying untreated sewage discharge points as major pollution sources for the Ganga River at Varanasi, serving as a powerful tool for pollution source identification. The study highlights the significance of augmenting water resources to mitigate pollution problems, as evidenced by the fluctuations in WQI values during the pre-monsoon, monsoon, and post-monsoon periods, with the sites S2, S8, and S9 identified as the most contaminated. The Ganga River in Varanasi is severely polluted because of the discharge of polluted water through various sources, resulting in poor water quality that is unsuitable for human consumption, including drinking and bathing at most of the ghats. The recommended approach for calculating an optimum WAWQI score is to consider TC, BOD/DO, and COD along with other relevant parameters. Overall, the water quality of the Ganga River is substandard and requires treatment prior to usage. The findings from this study can aid decision-makers in effectively managing and preserving the Ganga River in Varanasi.

The authors are thankful to Prof. P. K. Mishra, Department of Chemical Engineering IIT BHU, for providing laboratory facilities, and acknowledge Mr Saurabh Kumar Pathak and Mr Sanny Rathore, for support during sampling. The first author acknowledges the receipt of a fellowship from the University Grant Commission, Government of India. The corresponding author acknowledges the partial support from Banaras Hindu University through the Institute of Eminence (IOE) scheme.

Sample and data collection and analysis were carried out by Gurudatta Singh; Supriya Chaudhary and other authors contributed to writing; and Virendra Kumar Mishra contributed toward editing of the MS.

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

The authors declare there is no conflict.