Presence of heavy metals in surface metals is the mounting concern in the world due rapid industrialization and modernization of society, and surface water is being utilized both as source and sink for water consumption and wastewater discharge, respectively. The chemometric method were applied on river for pre and post monsoon seasons to determine the potential heavy metals and development of heavy metal pollution index (HPI). The cluster and factor analysis were applied on thirteen heavy metals monitored over eleven locations for characterizing the highly correlating and potential, respectively heavy metals. The proliferation of health risks of heavy metals was determined through cluster analysis and GIS. The Fe and Mn found exceeding the permissible limits in river and drain samples for both the pre-monsoon and post-monsoon season, whereas other heavy metals were found within the permissible limits throughout the study period. The study suggested that classification of pollutants and assessment of associated risk to human health could provide a valuable insight for the development of remediation measures to minimize the degradation of public health and appropriate treatment technology for minimizing the contamination of water sources.

  • River water quality is assessed using multivariate data analytics, GIS and HPI techniques.

  • Factor and cluster analysis derived the correlation and potential heavy metal affecting the water quality.

  • Health risk of heavy metals are assessed on human health.

  • Iron and manganese are found as potential and deteriorating heavy metals.

Water pollution is a growing cause of premature deaths and contributes annually to 1.7–2.2 million deaths around the world (Fuller et al. 2022). The primary concern for deaths is contaminated source of water followed by unavailability of sanitation facilities, where women and children are more prone to health risk exposure to polluted water compared to men (GBD 2019). The economic losses due to household water and sanitation services has decreased from 3.2 to 1% of India's gross domestic product (GDP), however, economic losses from the modern pollution areas such as industrial discharge of heavy metals, agricultural runoff, depleting groundwater etc. has increased from 0.75 to 1.1% of India's GDP from 2000 to 2019 (GBD 2019; World Bank 2022). The expansion of urban areas, industrial activities and socio-economic development of cities are the major source of modern water pollution (Herath et al. 2018; Kumar et al. 2022). The bioaccumulation of heavy metals in water, soil, crops, vegetables; long residence time in environment, non-biodegradability and toxicity make them more hazardous for terrestrial and aquatic life forms. Surface and groundwater are more susceptible to heavy metal pollution, especially in developing regions where untreated or partially treated wastewater from industrial areas and agricultural runoffs is discharged into open drains. These contaminants eventually flow into rivers, infiltrate the soil, and leach into groundwater, exacerbating pollution levels (Pandey & Kumari 2023).

Arid and semi-arid regions majorly rely on groundwater for domestic and commercial water supplies, and rural populations consume the water from the available sources without any treatment (Gupta et al. 2024). Therefore, it is essential to determine whether the water source is adequate for drinking, industrial, irrigation or other activities and the associated risk to the human health. Rivers in India face significant pollution, primarily from urban settlements, with cities located along their banks, contributing substantial volumes of wastewater. Singh et al. (2017) conducted a study to determine the heavy metal in soil quality from different areas using a geo-accumulation index and pollution index and found that agrochemical, vehicular emission and dumping of waste are the primary causes of heavy metal concentration. Kumar et al. (2020) determined the distribution trend of heavy metals in three rivers (Indus, Beas and Sutlej rivers). The study revealed Satluj River contained high concentrations of Pb (lead), Cu (copper) and Cr (chromium) compared to the Indus and Beas rivers, exceeding the permissible limits set by BIS drinking water quality standards. Herath et al. (2018) performed a study to determine the source of heavy metal contamination of Colombo Metropolitan Region and studied the soil samples, canal sediments, water samples and crops. The study found high concentrations of zinc, cobalt and lead in all the regions, dominantly caused by the anthropogenic sources. Adimalla (2020) assessed the concentration of heavy metals in 15 states of India and compared the results with the different geochemical guidelines and evaluated the hazard index of heavy metals. Kumar et al. (2019a, b) carried out a study of tubewell water to determine the trace metals and the source of metals was determined using multivariate and geospatial techniques. High concentrations of trace metals were reported near the industrial area. Akhtar et al. (2021) conducted a review on the impact of anthropogenic activities on the water resources. The anthropogenic sources were classified based on the land use land change (LULC) to determine the effluent discharge characteristics from each LULC type. The impact of anthropogenic activities was studied on the natural physiognomies such as soil, groundwater, surface water, aquatic environment and emerging inorganic pollutants. The study reveals the challenges faced by water resources and food production due to extensive urban development, and lack of sufficient infrastructure for handling wastewater load. Goyal et al. (2022) collected samples from pond sludge and groundwater for the quantification of the presence of nine trace metals (Pb, Cu, Ni, Zn, Cr, Cd, Fe, As, and Mn) and reported the ecological risk associated with the high concentration of trace metals where Pb, Cu, Ni, Fe and Mn were found to exceed the groundwater standards.

Most of the studies conducted in India were performed on Ganga river basin and southern plateaus which receive significant rainfall during monsoon, which helps wash away a significant portion of pollutants annually through precipitation. Some of the studies were carried out to identify the incidence of heavy metals in soil and sludge to determine their percolation in the food chain. However, no relevant study was found in the arid and semi-arid regions on the contamination of heavy metals in surface water source and associated health impacts. A comprehensive analysis of water bodies to identify the sources of heavy metals and assess their impact on human health from consuming contaminated water has not been conducted.

The spatial study of heavy metals by producing a heavy metal pollution index (HPI) is helpful in identifying and quantifying trends in water quality (Reza & Singh 2010; Prasad et al. 2020). It can also provide comprehensive information and assessments in a format that helps resource managers and regulatory agencies evaluate alternatives and make informed decisions. Moreover, cluster analysis (CA) and factor analysis (FA) were also performed to summarize information of large data sets as they offer better interpretation and understanding of water quality (Kannel et al. 2007; Arora & Keshari 2021). In particular, these FA and hierarchical cluster analysis (HCA) are increasingly in use for environmental studies, including measurement and monitoring of heavy metals in environmental media (Loska & Wiechuła 2003; Filgueiras et al. 2004; Ting et al. 2012; Zhao et al. 2017; Mishra et al. 2018).

This study was carried out on Mandakini River, Madhya Pradesh, India to assess the heavy metal contamination using HPI, GIS and statistical analysis techniques. The multivariate techniques and GIS were applied to meet the following objectives: 1) assessment of spatial distribution of heavy metals in Mandakini River, 2) determination of critical pollutants in the river and subsequent cause of pollution, 3) assessment of degree of contamination with heavy metals and 4) associated impact of heavy metals on human health. The study aims to provide new insights into regularly monitoring of water quality for heavy metals and their associated health impacts.

Study area

The Chitrakoot district is part of the Central Plateau and Hilly Bundelkhand region, spread over seven districts of Uttar Pradesh and six districts of Madhya Pradesh. The Chitrakoot district has red and black soil with a rocky and undulated surface. The climate is deficient in moisture content with semi-arid conditions. The climatic variation makes the study area vulnerable to different extreme weather events including heavy rainfall during monsoons and prolonged dry spells in subsequent seasons (Aslam & Parveen 2024). The heavy metals are monitored in Mandakini River at 11 stations in a 25 km stretch from the perennial origin of the stream to Karwi Bridge. The first nine stations are in Satna district of Madhya Pradesh, India and the last two sampling sites are in Chitrakoot district of Uttar Pradesh, India. Out of the 11 stations, nine samples were collected from the river and two samples were collected from drains. The first station was selected at Sati Anisuiya, which is considered a perennial river from this point onwards, where the river flows from south to north, as shown in Figure 1. The second station was selected at Sphatikshila, about 10 km downstream from Sati Anissuiya. The location is well known for Mahatma Gandhi Chitrakoot Gramodaya Vishwavidyalaya (MGCGV) and uses the Mandakini river water for drinking and other purposes. The third station was selected at Arogydham pool, about 0.5 km upstream of the confluence of Vaidehi Vatika drain with Mandakini River. This location could be useful for measuring the river's discharge, as it supplies drinking water to the town through the local administration of Chitrakoot. The fourth station was selected at Vaidehi Vatika Nallah before it meets the river Mandakini. The fifth station was selected at Jankikund about 0.5 km downstream of the point where Vaidehi Vatika Nallah meets the Mandakini river. This site is considered holy, and both locals and visitors come here daily for bathing. The sixth station was selected at Pramodvan about 2 km downstream of point where Vaidehi Vatika Nallah meets the river Mandakini. This location is situated halfway to the study area. This site is mostly used for bathing by devotees who come to Chitrakoot for eye surgeries. The seventh station was selected at Bharat Ghat (BG), located 0.5 km upstream from the confluence of the Paisuni drain and the Mandakini River. This site is used for boating by tourists and pilgrims. The eighth station was selected at Paisuni Nallah before it meets the main stream. The ninth station was selected at Ramghat about 0.5 km downstream of the point where Paisuni Nallah meets the stream. This is an important site for devotees of pilgrims and tourists for religious bathing. The 10th station was selected on Karwi bridge about 10 km downstream from the confluence of Paisuni Drain and Mandakini River as shown in Figure 2. The Karwi bridge connects the left and right sides of Karwi city. The 11th station was chosen downstream of Karwi bridge, where cremation and vegetation activities occur.
Figure 1

Study area of river Mandakini River.

Figure 1

Study area of river Mandakini River.

Close modal
Figure 2

Sampling locations along the drains and Mandakini River.

Figure 2

Sampling locations along the drains and Mandakini River.

Close modal

Data sampling

The water samples collected from 11 locations were divided into two categories to derive the variation in heavy metal concentration in the pre- and post-monsoon periods. The water sampling was performed for three years (2018, 2019 and 2021) in the months of April, May, September and October. The water samples collected during April and May are classified as pre-monsoon data and samples collected during September and October are taken as post-monsoon data. The location of sampling points along the drains and river are shown in Figure 2. The grab samples were collected at each point of the river from the centre of the stream or drain at about 0.6 m depth. Five hundred mL capacity polyethylene plastic bottles with a plastic cap were used for collecting samples for heavy metals parameters. Each sample bottle was washed three times before collecting water samples for trace metals. Firstly, every bottle was washed with 50% HNO3 acid and twice with nano pure distilled water. Samples were collected free from bottom sediment and away from stream banks by wading into the centre main current.

All the water samples were tested for heavy metals including arsenic (As) (μg/L), cadmium (Cd) (μg/L), chromium (Cr) (μg/L), copper (Cu) (μg/L), cobalt (Co) (μg/L), iron (Fe) (μg/L), manganese (Mn) (μg/L), nickel (Ni) (μg/L), lead (Pb) (μg/L), selenium (Se) (μg/L), vanadium (V) (μg/L), antimony (Sb) (μg/L) and zinc (Zn) (μg/L) using trace metal detection analytical techniques. Inductively coupled plasma mass spectrometry (ICP-MS) was used for analysis following the standard procedure prescribed by the American Public Health Association (APHA 2005). The concentration of heavy metals was compared with the BIS drinking standard 10500:2012 (BIS 2012), WHO Drinking Water Standard 2011 (WHO 2011) and USEPA Surface Water Standard. The most stringent permissible limit from these three standards was used for water quality analysis, as shown in Table 1. The referred permissible limit of heavy metals for assessment of water quality is highlighted with bold text in Table 1. For heavy metals where any agency does not prescribe any value, a zero concentration (considered acceptable in drinking water) is used to develop the Heavy Metal Pollution Index. All the parameters were tested from each sample, and testing was performed twice for the absent parameters to ensure no errors in the experimentation process. The experimentation results are categorized into pre- and post-monsoon scenarios.

Table 1

Trace metal standard acceptable limit for drinking and surface water along with method and instrumentation technique used in analysis

S. noParametersUnitBIS Drinking water, IS 10500: 2012WHO Drinking Water Standard 2011USEPA Surface Water StandardMethodInstrumentation technique
As μg/L 10 10 10 APHA-3125-B ICP-MS 
Cd μg/L 3 APHA-3125-B ICP-MS 
Cr μg/L 50 50 100 APHA-3125-B ICP-MS 
Cu μg/L 50 2,000 1,300 APHA-3125-B ICP-MS 
Fe μg/L 300 – 300 APHA-3125-B ICP-MS 
Mn μg/L 100 – 300 APHA-3125-B ICP-MS 
Ni μg/L 20 70 100 APHA-3125-B ICP-MS 
Pb μg/L 10 10 15 APHA-3125-B ICP-MS 
Zn μg/L 5,000 – 5,000 APHA-3125-B ICP-MS 
10 Co μg/L – – – APHA-3125-B ICP-MS 
11 Se μg/L 10 40 50 APHA-3125-B ICP-MS 
12 μg/L – – – APHA-3125-B ICP-MS 
13 Sb μg/L – 20 6 APHA-3125-B ICP-MS 
S. noParametersUnitBIS Drinking water, IS 10500: 2012WHO Drinking Water Standard 2011USEPA Surface Water StandardMethodInstrumentation technique
As μg/L 10 10 10 APHA-3125-B ICP-MS 
Cd μg/L 3 APHA-3125-B ICP-MS 
Cr μg/L 50 50 100 APHA-3125-B ICP-MS 
Cu μg/L 50 2,000 1,300 APHA-3125-B ICP-MS 
Fe μg/L 300 – 300 APHA-3125-B ICP-MS 
Mn μg/L 100 – 300 APHA-3125-B ICP-MS 
Ni μg/L 20 70 100 APHA-3125-B ICP-MS 
Pb μg/L 10 10 15 APHA-3125-B ICP-MS 
Zn μg/L 5,000 – 5,000 APHA-3125-B ICP-MS 
10 Co μg/L – – – APHA-3125-B ICP-MS 
11 Se μg/L 10 40 50 APHA-3125-B ICP-MS 
12 μg/L – – – APHA-3125-B ICP-MS 
13 Sb μg/L – 20 6 APHA-3125-B ICP-MS 

Methodology

The data generated from the testing of water samples were analyzed with spatial and temporal analysis techniques. Violin plots were developed for each heavy metal for both the pre- and post-monsoon seasons to determine the temporal variation in concentration of heavy metals. The spatial assessment of heavy metals in the study area is determined using a Geographical Information System (GIS). After the assessment of spatial and temporal variation in heavy metals, the multivariate statistical and HPI were applied to determine the critical pollutants and degree of pollution. Subsequently, health impacts were identified based on the identified heavy metal and pollution level of river. The graphical representation of the methodology of the study is shown in Figure 3.
Figure 3

Methodology to determine the water quality of Mandakini River.

Figure 3

Methodology to determine the water quality of Mandakini River.

Close modal

HPI

The concentration of heavy metals was tested using atomic absorption spectrophotometer (AAS) against the drinking water quality standards (BIS 10500:2012) in μg/L. The considerable difference in the acceptable and permissible limits of different heavy metals makes the water quality assessment a cumbersome process. To assess the degree of pollution in a water body due to the presence of heavy metals on a common scale, an HPI was implemented. The HPI deduces the water quality on a scale of 0–100, where 0 indicates no presence of any heavy metals and 100 represents the most hazardous conditions where all heavy metals were found to exceed the acceptable limit as per national standards. The identified heavy metals were weighted based on their significance on human health upon consumption and toxicity level (Kumar et al. 2019a, b). The sub-index of each heavy metal is calculated using Equation (1):
(1)
where is the sub-index for ith heavy metal, is the measured value of heavy metal and is the permissible limit of heavy metal. The summation of the weighted sub-index of heavy metals can be derived using Equation (2):
(2)

The HPI is sub-indexed into five categories and colour coded for easier evaluation of water quality, as shown in Table 2.

Table 2

HPI categories of water quality

HPI scoreCategoryColour coding
0–20 Excellent Green 
21–40 Good Yellow 
41–60 Moderate Orange 
61–80 Bad Red 
81–100 Severe Maroon 
HPI scoreCategoryColour coding
0–20 Excellent Green 
21–40 Good Yellow 
41–60 Moderate Orange 
61–80 Bad Red 
81–100 Severe Maroon 

Hierarchically aligned cluster analysis

Clustering is the method of grouping similar data points and each group is called a cluster. In hierarchically aligned cluster analysis (HACA), each data point defines the single membered cluster and combines to form the multi-membered cluster until the combination of all the data points results in a single large cluster (Shrestha & Kazama 2007). The distance between the data variables points is determined using Euclidean distance using the Equation (3):
(3)
where are the ith and jth observation of the kth variable. All the variables are considered as individual clusters and joined together based on the Euclidean distance. In between the cluster, the cluster distance is calculated using Ward's method, which calculates the smallest change in variance to merge the cluster. The cluster variance is determined using Equation (4):
(4)
where and are the centroid of clusters and , respectively and ‖‖ is the Euclidean distance between the centroid of cluster and . The pair of clusters with the smallest are merged to form a single cluster. The cluster variance is calculated until all the clusters join to form a single cluster. The sequence of merging of cluster is drawn using dendrogram plots. The optimum number of clusters are considered using mean Euclidean distance and mean variance of each cluster was measured using the elbow method.

Factor analysis

Factor analysis (FA), also known as data reduction technique, is widely used to reduce the dimensionality of data, pattern recognition and feature extraction. Factor analysis involves the mathematical procedure leading to reduction in the dimension of a large dataset without the loss of any significant information (Liu et al. 2003). The covariance between the variables and is calculated to derive the correlation between the variables, which quantifies the strength and direction of relationship. The correlation matrix is then decomposed to derive eigenvalues using Equation (5):
(5)
where R is the correlation matrix, Λ is the diagonal matrix of eigenvalues and Q is the eigenvalue matrix. The factor loading L from the eigenvalue is computed using Equation (6):
(6)
The factor to be retained is determined using the elbow method and thumb rule which specifies retaining factors with eigenvalues greater than 1. The results of both the methods are compared to retain the factors. The accepted factors are rotated orthogonally to achieve a more understandable and simplified structure, as shown in Equation (7):
(7)
(8)
where is the factor loading rate for variable j and factor k. The factor score is determined using Equation (6), where X is the standardized data matrix and W is the weighted matrix (Brown & Moore 2012).

The FA has the advantage of reducing dimension by observing the variance of components, eliminating less significant data and proving effective in handling large dimensional datasets. FA also works with non-parametric data and data pre-processing is not required for the identification of unique output.

Exploratory analysis

The violin plots of average concentration of heavy metals for both the identified seasons (pre- and post-monsoon) is illustrated in Figures 4 and 5, respectively, suggesting the variation in concentration of heavy metals. Further, the plots are divided into two parts based on the observed values of heavy metals at the sampling sites. Most of the heavy metals were found within the permissible limits for the pre- and post-monsoon periods including As, Cd, Cr, Cu, Ni, Pb, Zn, Se and Sb. The width of the violin plot represents the density of data points in a particular region. A wider section of a single violin indicates a higher probability of similar data values occurring in that range. The boundary of violin plots, known as kernel density estimation (KDE) boundary, develop over the distribution of data points. The shape of the KDE represents the distribution of data points over the interquartile range. The longer KDE boundary suggests the higher distribution and greater skewness in the data value.
Figure 4

Violin distribution of heavy metals during pre-monsoon period.

Figure 4

Violin distribution of heavy metals during pre-monsoon period.

Close modal
Figure 5

Violin distribution of heavy metals during post-monsoon period.

Figure 5

Violin distribution of heavy metals during post-monsoon period.

Close modal

Iron (Fe) was found to exceed the permissible limit at all sampling locations, while manganese (Mn) surpassed the permissible limit at a few locations, including Vaidehi Vatika Drain and Paisuni Drain, for both monitoring periods. As and Co had the shortest interquartile range during pre-monsoon and V during the post-monsoon period, showing no major variation in concentration of heavy metals spatially. The longest interquartile range with mediocre kernel density suggests significant variation in Zn concentration for pre and post-monsoon periods but remains within the permissible range. Similarly, multiple kernels were formed for Fe and Mn in both the seasons indicating the variation in concentration of variables at river and drain sampling locations. The lower kernel with greater width indicates the concentration of heavy metals at river sampling points, while the upper, narrower kernel represents the concentration at drain locations, which exceeds the permissible limits by multiple folds. The width of kernel density indicates that most of the river sampling sites maintain similar concentrations throughout the season. The upper kernel falls outside the violin plot for Mn, indicating it can be considered an outlier. The higher concentration of Fe and Mn illustrate the anthropogenic contamination of Paisuni drain might originate from discharge of effluents from corroded pipes, industrial effluent from metallurgical industries, etc.

The correlation plot, as shown in Figure 6, also suggests that Fe and Mn exhibit similar behaviour, with a high correlation coefficient of 0.96 during both seasons. Heavy metals with lower concentrations were also found to correlate with each other, illustrating a similar variation (low variation) among the sampling sites. The violin plot and correlation plot illustrate that most of the heavy metals could have a natural presence and remain in low and acceptable ranges, except Fe and Mn. The variation in Fe and Mn was also found to be highly correlated with V, followed by Ni, where Mn holds a stronger correlation compared to Fe for V and Ni, whereas Ni holds a strong correlation with Cr for both periods. The higher concentrations of Fe and Mn are likely generated from anthropogenic sources. Significant efforts are needed to reduce these concentrations, as the confluence of the Vaidehi Vatika Drain and Paisuni Drain is significantly impacting the river's water quality.
Figure 6

Correlation between the heavy metals at all the sampling sites.

Figure 6

Correlation between the heavy metals at all the sampling sites.

Close modal

HPI

The exploratory analysis suggests the considerable difference in the observed and acceptable concentration of heavy metals. Table 1 provides the permissible limits and Figures 4 and 5 provide observed concentrations during pre- and post-monsoon periods, respectively. The HPI was applied to evaluate the heavy metal contamination on a standardised scale, considering the difference in permissible limits of each heavy metal. The index ranges from 0 to 100, where 0 indicates no water source pollution, and 100 represents a highly polluted water source. Most of the sampling locations had excellent water quality, as HPI was found below 20, and few locations were found with HPI in the range of 21–23, as shown in Table 3. The water quality of Vaidehi Vatika Drain was categorized as ‘Good’ and does not significantly affect the water quality of the river. In contrast, the Paisuni Drain exhibited the most deteriorated water quality, classified under the ‘Severe’ and ‘Bad’ categories for pre- and post-monsoon periods, respectively. This indicates the substantial wastewater load carried by the Paisuni Drain from domestic and commercial areas. The water quality of the river downstream of Paisuni Drain falls into the ‘Good’ category, indicating that the impact of the drain is short-lived. This may be due to the significant flow in the river compared to the quantum of wastewater contributed by the Paisuni Drain. Tapping of drains from commercial activities could reform the water quality of the river to the ‘Excellent’ category.

Table 3

HPI of sampling site for pre- and post-monsoon period

S.No.Sampling locationHPI-PreHPI-Post
Sati Anusuiya Temple 17.76 19.71 
Sphtik Shila 23.05 20.47 
Arogyadham Ghat 18.61 17.06 
Vaidehi Vatika Drain 28.27 25.55 
Jankikund 20.42 18.93 
Pramodvan 19.16 19.89 
Bharat Ghat 16.81 16.07 
Paisuni Drain 82.24 75.54 
Ramghat 21.02 21.60 
10 Karwi Bridge upstream 11.35 11.59 
11 Karwi Bridge downstream 12.28 13.77 
S.No.Sampling locationHPI-PreHPI-Post
Sati Anusuiya Temple 17.76 19.71 
Sphtik Shila 23.05 20.47 
Arogyadham Ghat 18.61 17.06 
Vaidehi Vatika Drain 28.27 25.55 
Jankikund 20.42 18.93 
Pramodvan 19.16 19.89 
Bharat Ghat 16.81 16.07 
Paisuni Drain 82.24 75.54 
Ramghat 21.02 21.60 
10 Karwi Bridge upstream 11.35 11.59 
11 Karwi Bridge downstream 12.28 13.77 

Spatial and temporal variability

The geographical distribution of heavy metals along the river provide an insight on spatial distribution of heavy metals for both the seasons to record the temporal variation. The Cd, Se and Sb were observed as below detectable limits (BDL) as shown in Figure 7 for both the seasons, however Co and As were recorded as BDL during post-monsoon season and marginal concentration was observed during pre-monsoon at the Paisuni drain. Other heavy metals were found in the centre of the study area, where the Paisuni drain joins the river and deteriorates the water quality. The southern part of the study area was found to have the lowest concentration of heavy metals followed by the northern part. The higher concentrations were limited to the area surrounding the domestic area near Paisuni Drain. Habitation was observed between the Saphtik Shila sampling location and the area downstream of Ramghat. Throughout this stretch, the concentrations of Fe and Mn were consistently found to exceed the permissible limits. The high concentration indicates that domestic and industries activities are contributing to the degradation of river water quality.
Figure 7

The geographical distribution of heavy metals in the study area.

Figure 7

The geographical distribution of heavy metals in the study area.

Close modal

Hierarchical aligned cluster analysis (HACA)

HACA was performed to determine the similarity in the flux of heavy metals. The clusters were formed using Euclidean distance and Ward's method (Panda et al. 2006) was applied to calculate the distance between the clusters. The optimum number of clusters for analysis is obtained at mean Euclidean distance from the variables. The dendrograms are designed for pre- and post-monsoon seasons in the form of inverted trees, where each variable acts as a single cluster at the start of the clustering process and is merged with other clusters based on the minimum Euclidean distance. The process of clustering continues until a single cluster is formed from all the variables, as shown in Figure 8. During the pre-monsoon season, Mn and Co exhibit the highest similarity, forming a cluster at the shortest distance, followed by Fe, V, and As. Ni and Cr also form a separate cluster, and these two groups merge to create a single larger cluster. The cluster at the largest distance was formed by Cd and Pb, followed by Zn and Cu. The optimum number of clusters is considered at the mean distance of the variable at 2.49, as indicated by the red dashed line in Figure 8. During the pre-monsoon season, six clusters were identified at the mean distance. Cluster 1 consisted of six variables, Cluster 2 comprised two variables, and the remaining four clusters were each formed by a single variable.
Figure 8

Cluster formation of heavy metals at pre- and post-monsoon seasons.

Figure 8

Cluster formation of heavy metals at pre- and post-monsoon seasons.

Close modal

During the post-monsoon season, Mn and V formed a cluster at the shortest distance, followed by Fe. Ni and Cr formed another cluster, which later joined the cluster consisting of Mn, V, and Fe. Additionally, Zn and Pb formed a separate cluster. The mean distance was observed to be 3.17, resulting in four clusters. The largest cluster contained three variables, two clusters contained two variables each, and one cluster consisted of a single variable (Cu).

The shorter mean distance and larger clusters were formed during the pre-monsoon period compared to post-monsoon period suggesting that during pre-monsoon period, variables have greater similarity. The Ni and Cr form a single cluster in both the seasons but at a marginally greater distance than Mn, V and Fe. These five variables do not show considerable pattern variation in both the seasons, where As and Co are not found in post-monsoon season. The Zn forms a cluster with Cu in pre-monsoon whereas Zn forms a cluster with Pb in post-monsoon season, at roughly a similar distance from the centre in both the seasons. The comparison of dendrograms suggested that with the change in flow of river/drain, the pattern of flux in heavy metals do not change significantly. Both the seasons form clusters with similar heavy metals, however the mean distance increases with the increase in flow. The dilution and distribution of heavy metals with flow during post-monsoon produces some change in interaction of heavy metals.

Factor analysis

The analysis obtained after performing HACA suggested a similarity in the pattern of variation in heavy metals, however the influence in concentration of one heavy metal due to another metal is determined using factor analysis. The factor analysis divides the variables in multiple factors based on covariance of each variable and minimizes the dimensionality of data without losing any significant information. The first variance factor accounts for 57.85% of total variance in water quality of river during pre-monsoon season. The second and third variance factor contribute 27.04 and 15.11% of total variation, respectively. The cumulative variance of factors in the range of more than 0.85, 0.65–0.85 and less than 0.85 are considered as strong, moderate and weak correlation of variables, respectively. Co, Fe, Mn and V exhibited strong correlations during the pre-monsoon season, collectively contributing to factor 1, as shown in Figure 9. Ni was found to be moderately positively correlated with factor 1, whereas Cr and Cu showed moderate negative and positive correlation, respectively, contributing towards factor 2. Pb and Zn also indicated a moderate positive correlation contributed towards factor 2.
Figure 9

Factor loading for pre- and post-monsoon seasons.

Figure 9

Factor loading for pre- and post-monsoon seasons.

Close modal

The factor loading suggests different and weak correlation of variables during post-monsoon season compared to pre-monsoon season. At pre-monsoon season, the first two factors contribute towards 84.85% of total variance in water quality, whereas the first two factors lead to 79.25% of total variance, while factor 3 contributes the remaining 20.74%, as shown in Figure 9. Mn, Fe, and V exhibited a strong positive correlation with Factor 1, while Ni showed a moderate positive correlation. Pb and Zn displayed a strong and moderate positive correlation, respectively, with Factor 2. Cu contributed to Factor 3 with a moderate positive correlation, whereas Cr and Ni demonstrated a moderate negative correlation with Factor 2. Consistent with the cluster analysis results, the variables showed variance across different factors. During the pre-monsoon season, Fe, Mn, As, V, and Co formed a single cluster at the mean distance, contributing to Factor 1. In the post-monsoon season, Fe, Mn, and V formed a single cluster at the mean distance and strongly correlated with Factor 1.

Assessment of critical parameters

The concurrence in the variables appearing in clusters and factor loading distribution pattern suggest that similar parameters are influencing the water quality significantly and have similar patterns of variation among both seasons. The Mn, Fe, As, V and Co were found as principal variables during pre-monsoon, and Mn, Fe and V were the principal variables at post-monsoon season, where As and Co were found below detectable limits. The results indicate that there are common sets of heavy metals affecting the water of river dominantly originating from Paisuni drain receiving the wastewater load from anthropogenic activities (Zhou et al. 2020; Mishra et al. 2021). The heavy metals found in lower concentrations throughout the study area are likely generated from natural sources, as these lower concentrations are observed in uninhabited areas. The temporal distribution of heavy metals suggests that Fe and Mn remain above the permissible limits throughout the year as per IS standards for drinking water. The monsoon has not significantly reduced the concentration of heavy metals, as anthropogenic activities contribute pollutants to the drain throughout the year. The spatial distribution maps suggest that the highest concentrations of Fe and Mn were found near the centre of the city, where Paisuni drain joins the river, and confirms the role of the drain in deteriorating the water quality of the river (Mishra et al. 2021). The factor analysis and clustering provide significant information about the presence and interaction of heavy metals that can be applied by biologists to determine the relevant heavy metals required to be estimated at different sampling sites and policy makers may utilize the findings to identify appropriate treatment techniques for the control and prevention of water quality deterioration.

Assessment of health risk

The presence of heavy metals in water sources beyond permissible limits may affect the mental and physical growth of humans and cause detrimental damage to the human body including the nervous system, cancer and even death due to prolonged consumption. The consumption of heavy metals such as As, Cd, Co, Ni, Cr, Pb and Zn have high toxicity levels and could cause skin manifestations, kidney disease, vascular disease (Rehman et al. 2018), liver damage, neurological signs, damage the fetal brain and diseases of the kidneys, circulatory system, and nervous system (Fallahzadeh et al. 2017). The quantified concentration of all the heavy metals was found to be below the permissible limits except for Fe and Mn. Fe was found above 1,000 μg/L against the permissible limit of 300 μg/L, whereas Mn was found up to 400 μg/L against the acceptable limits of 100 μg/L as per IS standards. The consumption of water with high concentrations of iron and manganese pose several health risks and practical issues. The consumption of water with high concentrations of Fe can cause nausea, stomach pain, vomiting, and constipation, however prolonged consumption may affect the internal organs, particularly the liver and heart (Fallahzadeh et al. 2017). Genetic disorders also allow the accumulation of iron in the body and lead to joint pain and fatigue (Abbaspour et al. 2014). Chronic exposure of manganese is more detrimental and leads to neurological disorders including Parkinson disease, muscle spasm, difficulty in walking etc. Infants and children under the age of five are at high risk of Mn toxicity, which can impair cognitive function and lower IQ development (Gražulevičienė & Balčius 2009). Studies have also shown excessive Mn intake can negatively affect reproductive growth (Studer et al. 2022). The consumption of water contaminated with Fe leads to liver cirrhosis if consumed above 1 mg/L and also induces chronic diseases such as heart disease and diabetes. While iron primarily affects gastrointestinal health and can contribute to microbial growth in water systems, manganese exposure is linked to neurological and developmental effects, particularly in children. Additionally, both metals cause significant aesthetic and maintenance problems in water supplies. It is crucial to monitor and manage the levels of these metals in drinking water to protect public health and maintain water quality.

Multivariate statistical techniques and GIS were applied to determine spatial and temporal variation, respectively. The critical heavy metals affecting the water quality of Mandakini River, Madhya Pradesh, India are being determined using factor analysis and overall water quality is evaluated using HPI. Violin plots have been used for the first time for the assessment of temporal variation in heavy metal concentration over different seasons. GIS plots were constructed to monitor the pre- and post-monsoon spatial variation in water quality. Both the pre-and post-monsoon seasons showed that iron and manganese remain above the permissible limits, whereas other parameters remain within the acceptable limit and were found in low concentrations throughout the year. The monsoon reduces the concentration of Fe and Mn slightly, but not enough to make it acceptable for any use. The Paisuni Drain, which receives wastewater from anthropogenic activities, contributes a significant concentration of Fe and Mn. Cluster analysis indicated that Fe, Mn, and V, followed by Ni and Cr, exhibit similar flux patterns. The factor analysis also found similar variables affecting the water quality considerably. The principal factors comprising Fe, Mn, V, As, and Co contribute strongly to affecting water quality. Similar variables in both analyses suggest that the variables with the most identical variations are also critical in affecting water quality, as they exhibit the highest concentrations. Integrating innovative approaches combining statistical, geographical methods, and indices provide a holistic methodology for effectively assessing water quality. The control and preventive measures on Fe and Mn could resolve water quality challenges. Policymakers can adopt a similar methodology to identify and monitor critical variables that need to be managed to improve water quality effectively.

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

The authors declare there is no conflict.

Abbaspour
N.
,
Hurrell
R.
&
Kelishadi
R.
(
2014
)
Review on iron and its importance for human health
,
Journal of Research in Medical Sciences: The Official Journal of Isfahan University of Medical Sciences
,
19
(
2
),
164
.
Akhtar
N.
,
Syakir Ishak
M. I.
,
Bhawani
S. A.
&
Umar
K.
(
2021
)
Various natural and anthropogenic factors responsible for water quality degradation: a review
,
Water
,
13
(
19
),
2660
.
APHA
(
2005
)
Standard Methods of Water and Wastewater
, 21st edn.
Washington, DC
:
American Public Health Association
, pp.
2
61
.
Aslam
A.
,
Parveen
A.
, (
2024
)
Monitoring spatio-temporal pattern of meteorological drought stress using Standardized Precipitation Index (SPI) over Bundelkhand Region of Uttar Pradesh, India
. In:
Singh
A.
,
Jamal
S.
&
Ahmad
W. S.
(eds.)
Climate Change, Vulnerabilities and Adaptation: Understanding and Addressing Threats with Insights for Policy and Practice
.
Cham, Switzerland
:
Springer Nature
, pp.
203
214
.
Brown
T. A.
&
Moore
M. T.
, (
2012
)
Confirmatory factor analysis
. In:
Hoyle
R. H.
(ed.)
Handbook of Structural Equation Modeling
.
New York
:
Guilford Publications
, pp.
361
379
.
Bureau of Indian Standards (BIS)
(
2012
)
Bureau of Indian Standard Drinking Water Specification (Second Revision) IS: 10500:2012
.
New Delhi
:
Bureau of Indian Standards
.
Fallahzadeh
R. A.
,
Ghaneian
M. T.
,
Miri
M.
&
Dashti
M. M.
(
2017
)
Spatial analysis and health risk assessment of heavy metals concentration in drinking water resources
,
Environmental Science and Pollution Research
,
24
,
24790
24802
.
Fuller
R.
,
Landrigan
P. J.
,
Balakrishnan
K.
,
Bathan
G.
,
Bose-O'Reilly
S.
,
Brauer
M.
,
Caravanos
J.
,
Chiles
T.
,
Cohen
A.
,
Corra
L.
&
Cropper
M.
(
2022
)
Pollution and health: a progress update
,
The Lancet Planetary Health
,
6
(
6
),
e535
e547
.
GBD, Diseases and Injuries Collaborators
(
2019
)
Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease study 2019
,
Lancet
,
396
,
1204
1222
.
Goyal
V. C.
,
Singh
O.
,
Singh
R.
,
Chhoden
K.
&
Malyan
S. K.
(
2022
)
Appraisal of heavy metal pollution in the water resources of Western Uttar Pradesh, India and associated risks
,
Environmental Advances
,
8
,
100230
.
Gražulevičienė
R.
&
Balčius
G.
(
2009
)
Assessment of iron and manganese concentration changes in kaunas city drinking water distribution system
,
Environmental Research, Engineering and Management
,
50
(
4
),
37
43
.
Gupta
L. N.
,
Arora
S.
,
Ahirwar
M. K.
,
Pandey
A. P.
,
Khare
A. B.
&
Gupta
G. S.
(
2024
)
Assessment of surface and groundwater quality in Tikamgarh City, Madhya Pradesh, India
,
The Holistic Approach to Environment
,
14
(
2
),
40
48
.
Herath
D.
,
Pitawala
A.
,
Gunatilake
J.
&
Iqbal
M. C. M.
(
2018
)
Using multiple methods to assess heavy metal pollution in an urban city
,
Environmental Monitoring and Assessment
,
190
,
1
15
.
Kannel
P. R.
,
Lee
S.
,
Kanel
S. R.
,
Khan
S. P.
&
Lee
Y. S.
(
2007
)
Spatial–temporal variation and comparative assessment of water qualities of urban river system: a case study of the river Bagmati (Nepal)
,
Environmental Monitoring and Assessment
,
129
,
433
459
.
Kumar
M.
,
Nagdev
R.
,
Tripathi
R.
,
Singh
V. B.
,
Ranjan
P.
,
Soheb
M.
&
Ramanathan
A. L.
(
2019a
)
Geospatial and multivariate analysis of trace metals in tubewell water using for drinking purpose in the upper Gangetic basin, India: heavy metal pollution index
,
Groundwater for Sustainable Development
,
8
,
122
133
.
Kumar
V.
,
Parihar
R. D.
,
Sharma
A.
,
Bakshi
P.
,
Sidhu
G. P. S.
,
Bali
A. S.
,
Karaouzas
I.
,
Bhardwaj
R.
,
Thukral
A. K.
,
Gyasi-Agyei
Y.
&
Rodrigo-Comino
J.
(
2019b
)
Global evaluation of heavy metal content in surface water bodies: a meta-analysis using heavy metal pollution indices and multivariate statistical analyses
,
Chemosphere
,
236
,
124364
.
Kumar
V.
,
Sharma
A.
,
Kumar
R.
,
Bhardwaj
R.
,
Kumar Thukral
A.
&
Rodrigo-Comino
J.
(
2020
)
Assessment of heavy-metal pollution in three different Indian water bodies by combination of multivariate analysis and water pollution indices
,
Human and Ecological Risk Assessment: An International Journal
,
26
(
1
),
1
16
.
Kumar
P.
,
Mishra
V.
,
Yadav
S.
,
Yadav
A.
,
Garg
S.
,
Poria
P.
,
Farooqi
F.
,
Dumée
L. F.
&
Sharma
R. S.
(
2022
)
Heavy metal pollution and risks in a highly polluted and populated Indian river–city pair using the systems approach
,
Environmental Science and Pollution Research
,
29
(
40
),
60212
60231
.
Liu
C. W.
,
Lin
K. H.
&
Kuo
Y. M.
(
2003
)
Application of factor analysis in the assessment of groundwater quality in a blackfoot disease area in Taiwan
,
Science of the Total Environment
,
313
(
1–3
),
77
89
.
Mishra
S.
,
Kumar
A.
,
Yadav
S.
&
Singhal
M. K.
(
2018
)
Assessment of heavy metal contamination in water of Kali river using principle component and cluster analysis, India
,
Sustainable Water Resources Management
,
4
,
573
581
.
Panda
U. C.
,
Sundaray
S. K.
,
Rath
P.
,
Nayak
B. B.
&
Bhatta
D.
(
2006
)
Application of factor and cluster analysis for characterization of river and estuarine water systems–a case study: Mahanadi River (India)
.
Journal of Hydrology
,
331
(
3–4
),
434
445
.
Pandey
S.
,
Kumari
N.
, (
2023
)
Impact assessment of heavy metal pollution in surface water bodies
. In:
Shukla
S. K.
,
Kumar
S.
,
Madhav
S.
&
Mishra
P. K.
(eds.)
Metals in Water: Global Sources, Significance, and Treatment
.
Amsterdam
:
Elsevier BV
, pp.
129
154
.
Prasad
B.
,
Soni
A. K.
,
Vishwakarma
A.
,
Trivedi
R.
&
Singh
K. K. K.
(
2020
)
Evaluation of water quality near the Malanjhkhand copper mines, India, by use of multivariate analysis and a metal pollution index
,
Environmental Earth Sciences
,
79
,
1
23
.
Rehman
K.
,
Fatima
F.
,
Waheed
I.
&
Akash
M. S. H.
(
2018
)
Prevalence of exposure of heavy metals and their impact on health consequences
,
Journal of Cellular Biochemistry
,
119
(
1
),
157
184
.
Reza
R.
&
Singh
G.
(
2010
)
Heavy metal contamination and its indexing approach for river water
,
International Journal of Environmental Science & Technology
,
7
,
785
792
.
Studer
J. M.
,
Schweer
W. P.
,
Gabler
N. K.
&
Ross
J. W.
(
2022
)
Functions of manganese in reproduction
,
Animal Reproduction Science
,
238
,
106924
.
The World Bank
(
2022
)
World Development Indicators
.
Washington, DC
:
The World Bank
. .
Ting
Y. S.
,
Freeman
K. C.
,
Kobayashi
C.
,
De Silva
G. M.
&
Bland-Hawthorn
J.
(
2012
)
Principal component analysis on chemical abundances spaces
,
Monthly Notices of the Royal Astronomical Society
,
421
(
2
),
1231
1255
.
World Health Organization (WHO)
(
2011
)
Guidelines for Drinking-Water Quality
, 4th edn.
Geneva
:
World Health Organization
, pp.
8
18
.
Zhao
G.
,
Ye
S.
,
Yuan
H.
,
Ding
X.
&
Wang
J.
(
2017
)
Surface sediment properties and heavy metal pollution assessment in the Pearl River Estuary, China
,
Environmental Science and Pollution Research
,
24
(
3
),
2966
2979
.
Zhou
Q.
,
Yang
N.
,
Li
Y.
,
Ren
B.
,
Ding
X.
,
Bian
H.
&
Yao
X.
(
2020
)
Total concentrations and sources of heavy metal pollution in global river and lake water bodies from 1972 to 2017
,
Global Ecology and Conservation
,
22
,
e00925
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).