Mapping drinking water contamination in Gazipur City restaurants using WQI and multivariate analysis Open Access

Groundwater serves as a primary source of drinking water for domestic, industrial, and agricultural purposes in many countries, including Bangladesh, and is crucial to sustaining populations worldwide (UNESCO 2021). In urban areas, approximately 66% of households (with a range of 17% to 93% across individual countries) and 60% of rural households (with a range of 22–95%) rely on groundwater for drinking (Carrard et al. 2019). However, when the quality of drinking water does not meet established international and national guideline values, such as those set by the World Health Organization (WHO 2022) or the Environmental Conservation Rules (ECR 2023), it can lead to significant adverse effects on human health. Traditionally, the suitability of drinking water is determined by analyzing a range of physical, chemical, and microbiological parameters and comparing the results to regulatory standards (Ameen 2019). While this approach ensures legal compliance, interpreting test results across multiple parameters and assessing overall water quality can be a complex task. To streamline and simplify this process, water quality indices (WQIs) have been widely adopted. These indices condense extensive data on drinking water quality into a single, easily interpretable number, offering a comprehensive representation of water quality while maintaining scientific rigor (Menniti & Guida 2020; Uddin et al. 2021, 2022; Fatima et al. 2022).

In recent years, WQI has become a common tool for classifying and characterizing water resources for various purposes, including drinking, agricultural, surface water, and industrial uses (Yadav et al. 2010; Udeshani et al. 2020; Khan et al. 2023; Nsabimana & Li 2023). Numerous countries, including Bangladesh (Akter et al. 2016; Rahaman et al. 2019), India (Yadav et al. 2018; Khangembam & Kshetrimayum 2019; Mukate et al. 2019; Banerjee et al. 2024), Saudi Arabia (Al-Omran et al. 2015), Tunisia (Ketata et al. 2012), Ecuador (Roldán-Reascos et al. 2024), China (Wu et al. 2020), and Pakistan (Solangi et al. 2020; Saleem et al. 2024; Ullah et al. 2024) have applied WQI to assess water quality for drinking purposes. Various WQI have been developed worldwide, including the US National Sanitation Foundation Water Quality Index (NSFWQI) (Brown et al. 1970), the Canadian Council of Ministers of the Environment Water Quality Index (CCMEWQI) (Khan et al. 2003), the British Columbia Water Quality Index (BCWQI), and the Oregon Water Quality Index (OWQI) (Debels et al. 2005; Kannel et al. 2007).

In addition to WQI, multivariate statistical techniques such as principal component analysis (PCA) and factor analysis (FA) are frequently used to assess the quality of surface and groundwater (Deng & Liu 2020). These methods help identify specific contaminants and their sources, providing insight into the environmental processes affecting water quality (Liu et al. 2003; Shrestha & Kazama 2007; Bu et al. 2009; Varol & Davraz 2014; Howladar et al. 2017, 2021; Dubrovskaya 2023). For example, Bu et al. (2009) used hierarchical cluster analysis (HCA) to assess contamination levels in the Jinshui River, China, classifying 12 sampling spots into three clusters based on 22 variables. They also used FA, which revealed five components explaining 90.01% of the overall variance. Similarly, Shrestha & Kazama (2007) grouped 13 sampling spots into three clusters based on water quality, using FA and PCA to identify latent factors explaining 65.39, 73.18, and 77.61% of the overall variance in terms of pollution levels. These methods are invaluable for understanding hydro-chemical processes and contamination sources (Kumar et al. 2006).

Given the growing environmental and public health concerns, Pearson correlation analysis is often employed to establish relationships between physicochemical parameters and to trace the origins of contaminants. For instance, studies have used PCA and FA to assess groundwater pollution in areas prone to health issues such as black foot disease in Taiwan (Liu et al. 2003) and other contaminated regions like Dinajpur, Bangladesh (Howladar et al. 2017). These analyses help identify the most significant factors influencing water quality, such as industrial effluents, agricultural runoff, and natural geological processes.

Many studies have utilized WQI and multivariate analysis to assess water quality. A recent study by Howladar et al. (2021) demonstrated the utility of these methods in identifying contaminants like turbidity, dissolved oxygen (DO), total dissolved solids (TDS), hardness, chemical oxygen demand (COD), total suspended solids (TSS), and microbial contamination. Their findings highlighted the poor water quality in many regions, largely due to improper management of waste and drainage systems. Rahaman et al. (2019) used cluster analysis and WQI to assess water quality in Rajshahi City, Bangladesh, identifying different water types based on chemical compositions. Similarly, studies like those by Akter et al. (2016) have shown how groundwater contamination, particularly from heavy metals such as arsenic, manganese, and iron, contributes to the decline in water quality.

Bangladesh faces significant challenges in managing groundwater quality due to contamination from industrial effluents, agricultural runoff, and high population density in urban centers like Gazipur (Hossain & Rahman 2018). The UNESCO World Water Assessment Programme (2024) report underscores the continuing reliance on groundwater, exacerbating the risk of water contamination. To address this, Bangladesh has implemented various programs aimed at improving the health of vulnerable communities and promoting access to clean water. Previous studies have focused on tube well water and water-related health issues, but no comprehensive studies have been conducted on water quality in Gazipur City using WQI and multivariate analysis to identify contamination zones.

Gazipur, an industrial hub and one of the most densely populated areas in Bangladesh, is experiencing increasing demand for groundwater, which has led to significant depletion and contamination of water sources (Aziz & Sulaiman 2022; Rana & Moniruzzaman 2024). Since 2003, population and industrial activities in the city have rapidly increased, impacting groundwater levels (Parvin 2019). According to Banglapedia (2001), approximately 85.62% of the population relies on tube wells, with the remainder using other sources. Many people, particularly in restaurants and tea stalls, rely on groundwater or locally sourced jar water, which is prone to contamination. Studies such as that by Sarker et al. (2016) in Sylhet City have revealed high levels of fecal coliform and excessive iron in water used in restaurants, highlighting the public health risks associated with unsafe drinking water practices. The uncontrolled extraction of groundwater to meet demand poses serious long-term health risks, as customers often consume water unaware of its contamination (Chandnani et al. 2022).

To date, no comprehensive study has evaluated groundwater quality specifically in restaurant settings within Gazipur City using an integrated approach combining multiple WQI methods and advanced multivariate analyses. The novelty of this research lies in the simultaneous application of integrated WQI (IWQI), assigned weight WQI (AWWQI), and weighted arithmetic WQI (WAWQI), combined with hierarchical clustering, Pearson correlation, FA, and microbial assessments, to provide an in-depth understanding of groundwater contamination. Specifically, this study aims to: (1) evaluate physicochemical and microbial water quality parameters collected in 2019 from restaurants across Gazipur City; (2) identify contamination zones using hierarchical clustering; and (3) investigate pollution sources through correlation and factor analyses. The findings will support targeted water quality management and policy development to safeguard public health in rapidly urbanizing regions.

The water used for drinking purposes was collected in 2019 from 173 restaurants and tea stalls in 18 regions of the study area, as shown in Figure 1 and described in Karim et al. (2023). These drinking water samples were tested for 17 different water quality parameters, and their measurement procedures are also described in Karim et al. (2023). The results of the analysis are presented in Table 1.

The linkage distance between different groups then determines the variation of water quality with respect to space and represents it in a dendrogram. In this study, HCA has been done using IBM SPSS 25.

This study applies three WQI methods – the WAWQI, AWWQI, and IWQI, to comprehensively evaluate drinking water suitability from multiple analytical perspectives. The analysis is based on water standards set by The Environment Conservation Rules, 2023 (ECR 2023). The lower and upper values of these standards are defined as desirable limits (DL) and permissible limits (PL), respectively. The WAWQI method integrates physicochemical parameters into a weighted mathematical equation, emphasizing the relative significance of each parameter in water quality assessment (Călmuc et al. 2018). The AWWQI assigns parameter-specific weights based on their local importance, providing context-sensitive insights into water quality conditions (Brown et al. 1970). The IWQI method offers a comprehensive and unbiased evaluation by considering both the DL and PL threshold limits, aligned with established drinking water quality standards (Mukate et al. 2019). The combined use of these methods enables a robust, nuanced assessment of groundwater quality within the study area.

The WAWQI method has been used extensively all around the world. In this study, we followed (Yadav et al. 2010; Howladar et al. 2021) to calculate the WQI. Initially, we calculate the subindex value using Equation (3).

where k is the proportionality constant and ⁠.

Table A1 (in Supplementary Materials) shows the different parameters weightage values according to the Bangladesh guideline ECR 2023 for calculating the WQI of the dataset.

The scale to measure the WQI is shown in Table A2 (in Supplementary Materials), and using this scale, we measure the quality of drinking water in this study.

These range values are represented in Table A3 (in Supplementary Materials) following the ECR 2023 guideline of Bangladesh.

Sub index (SI) computations are applied here to find concentrations of parameters that are either below or above DL and PL, respectively, which will heavily deteriorate the quality of drinking water. On the other hand, concentrations of parameters found between DL and MPL are considered excellent for drinking water quality, as shown in Figure A1 (in Supplementary Materials).

Finally, the drinking water quality rating is evaluated through the IWQI values according to Table A5 (in Supplementary Materials).

In this WQI analysis, firstly, weights are assigned to the 13 selected parameters. The weights range from 1 to 5 based on their importance. However, some of the weights are assigned based on previous studies (Kumar 2004; Ramakrishnaiah et al. 2009; Yadav et al. 2010). The mean values are taken as the assigned weight. Moreover, the remaining weights are assigned based on their role in influencing drinking water quality. An assigned weight of 5 indicates the most significant parameter, while the lowest value of 1 indicates the least significant parameter.

The assigned and relative weight values used in the AWWQI calculation are presented in Table A6 (Supplementary Materials), and were determined based on Bangladesh ECR 2023 guidelines, as well as established methodologies from Ramakrishnaiah et al. (2009) and Kumar (2004), to reflect the relative significance of each parameter in groundwater quality assessment.

Now there are several scales available to rate or classify drinking water quality based on the WQI value. These studies are shown in Table A7 (in Supplementary Materials), and we have followed the scaley by Yadav et al. (2010) because it is the most recent one.

The Pearson coefficient of correlation, denoted by r, is a numerical measure of the correlation between two correlated variables or pairs of variables (Howladar et al. 2021). It assesses the strength, association and direction of the linear relationship that exists between these variables. R programming is used to carry out the Pearson correlation matrix analysis.

In the study area, variation in physico-chemical parameters in drinking water infer a significant impact on the WQI, and multivariate analysis identifies the sources of pollutants that are potentially harmful to consumers. The three WQI methods involve categorization and the identification of pollutant sources that may cause health problems.

Similarly, Turbidity, DO, TDS, EC, Hardness, and Manganese also show higher values than Cluster I and Cluster II. Unlike Chlorine, Fluoride and Sulfate have higher values in Cluster II than in Cluster I and Cluster III. However, the values of Color, Iron, Nitrate, and Ammonia in Cluster I are more significant than in Cluster II and Cluster III. From Table 2, we can draw the following conclusions: Cluster I is a type of drinking water characterized by Color, Iron, Nitrate, and Ammonia; Cluster II is a type of drinking water characterized by Chlorine, Fluoride, and Sulfate; and Cluster III is a type of drinking water characterized by Hardness, pH, TDS, DO, and Turbidity. In the study area, these parameters dominate groundwater quality in the vicinity of restaurants and tea stalls.

Drinking WQIs have been calculated using three methods, and the output of the observations is illustrated. Different methods show different results. There will be a discussion of the outcomes from various methods in this section.

Also, 9 and 13% of the total sample's drinking water quality are in the very poor and poor categories. Therefore, according to this method, the overall drinking water quality of Cluster I can be considered an unsuitable category. In Cluster II, the results show that the drinking water quality of 38% of the total samples is unsuitable, whereas 29% of the total samples are in the excellent category. However, it also shows that the drinking water quality in 14, 10, and 10% of the total samples is in the good, poor, and very poor categories, respectively. Therefore, according to this method, the highest 38% of samples are in the unsuitable category, so we can conclude that the drinking water quality of Cluster II can be considered unsuitable overall. In Cluster III, this method indicates that the drinking water quality of 57% of the total samples is in the unsuitable category, whereas only 10% of the total samples are an excellent category. It also shows that only 13% of samples are in the good quality category, while 11 and 10% of samples are in the poor and very poor quality categories, respectively. So, Cluster III can also be considered an unsuitable category, considering drinking water quality according to this method.

Under the IWQI method, Cluster I can be considered a poor category overall. However, considering the IWQI method shown, Cluster II represents that the drinking water quality of 19% of samples in an unsuitable category, but there are no samples in the excellent category in Cluster II. Therefore, a maximum of 38% of samples is acceptable, whereas 19% is in the unsuitable category. However, again, there are no samples in the good category in Cluster II. Therefore, the drinking water quality of Cluster II can be considered in the acceptable category. In Cluster III, the drinking water quality of 42% of samples is in the unsuitable category, whereas only 2% of samples are in the excellent category. IWQI also shows that the drinking water quality of 7, 18, and 31% are in the good, acceptable, and unsuitable categories. Therefore, the drinking water quality of Cluster III can be considered in the unsuitable category.

As 49% of the samples were in the good quality category, Cluster I can be considered to have good drinking water quality according to this method. This method indicates that there are no unsuitable samples in Cluster II, while 10% are rated as excellent. Among the 43% of samples in Cluster II, the majority fall into the good category, but 29 and 19% are classified as poor and very poor, respectively. Therefore, we can conclude that Cluster II can be considered to have good drinking water quality. Moving on to Cluster III, this method reveals that only 1% of the samples are unsuitable, with none falling into the excellent category. 27, 52, and 19% of the samples in Cluster III are categorized as good, poor, and very poor, respectively. Thus, Cluster III can be considered to have poor drinking water quality.

According to the overall scenario in Figure 6, about 1, 49, and 31% of the sample's overall drinking water quality falls into the unsuitable category, as shown by the AWWQI, WAWQI, and IWQI methods, respectively. These results reveal that WAWQI and IWQI give similar results, whereas AWWQI differs considerably. According to the AWWQI, the drinking water quality in Gazipur City is in the good category. In contrast, WAWQI and IWQI show that the drinking water quality of Gazipur City is in an unsuitable category. The discrepancies in the results between the AWQI and other WQIs are due to the weighted values assigned in the AWQI calculation, which depend on the influence of specific parameters on drinking water quality. The AWWQI method involves assigning weights based on literature and expert opinion, introducing potential biases and arbitrary values that can significantly alter WQI results (Amuah & Kpiebaya 2022). The WAWQI model may experience an eclipsing effect, where suboptimal aggregation and parameter selection lead to ambiguity in the interpretation of water quality (Smith 1990). While the IWQI model effectively integrates multiple parameters using desirable and PLs, it has limitations in scenarios where DLs are not explicitly defined, potentially affecting clarity (Bureau of Indian Standards 2012; Rajkumar et al. 2022). Thus, despite these limitations, the IWQI is recommended for Gazipur City due to its lower bias and direct incorporation of health-based guidelines. Moreover, all three WQIs use different scales and equations to calculate the values of the WQI. This variation in methodology could also contribute to differences in results among the indices. Also, 1, 13, and 2% of the drinking water quality are in the excellent category according to AWWQI, WAWQI, and IWQI, respectively. Since the results of both WAWQI and IWQI show slightly similar outcomes (based on their rating of WQI value) in assessing the quality of drinking water, the accuracy of AWQI is low in comparison to other WQIs. Moreover, studies conducted by Mou et al. (2023) and Shaibur et al. (2021) also found unsuitable water quality in Jashore and Khulna, Bangladesh. In Mou et al.'s (2023) study, it was discovered that approximately 40% of the physicochemical parameters exceeded the PLs for drinking water. Furthermore, Shaibur et al. (2021) also reported similar findings, indicating that 46% of the water samples exceeded the standard limits, with elevated levels of Ca, Mg, and Fe surpassing the WHO standards. In a study conducted by Mahmud et al. (2020) to assess the groundwater quality of Khulna city using WQI, it was found that 40% of the total samples needed to be pre-treated before being used for drinking purposes. Similarly, in another study conducted in Rajshahi city of Bangladesh by Rahaman et al. (2019), it was found that during the pre-monsoon season, Ca, Mg, Fe, Cl, and Mn levels were above the ECR guidelines, accounting for about 45% of the samples, rendering the groundwater unsuitable for drinking. They concluded that anthropogenic activities, weathering of calcareous and clay minerals, and municipal sewage are the main sources of groundwater pollution. Rio et al. (2022) also reported excessive untreated Pb, Fe, and Mn concentrations from textile and dyeing effluent exceeding guideline limits in Bangladesh. Moreover, as an industrial hub, Gazipur City's groundwater contamination primarily arises from industrial effluents and agricultural runoff. Textile and dyeing effluents contain high dissolved salts and dye chemicals, significantly increasing TDS in nearby groundwater (Rabbi et al. 2016). Pharmaceuticals also contribute substantially, accounting for approximately 16% of Bangladesh's industrial effluent load (Hasan et al. 2019). Heavy metals such as Fe and Mn from agricultural runoff (Sahen et al. 2025), along with excessive nitrate levels from fertilizers, further degrade groundwater quality, with fertilizers alone responsible for 64% of nitrate pollution nationwide (Alam et al. 2024). This type of drinking water contamination poses a huge risk to public health. Thus, the WQI's outcome of this study will create awareness among the public and the authorities to treat the drinking water of restaurants and tea stalls.

The 173 samples of 13 variables were collected from the study area, and among all the samples, the most prevalent parameters were assessed by the FA method. The result of FA in Cluster I, Cluster II and Cluster III, total cumulative variance, which is explained respectively by 73.095%, 91.335%, and 68.013%.

Table 3 shows that Factor 1 has a strong positive loading value of TDS, EC, and Hardness and a moderate loading value of DO. In Cluster II, factor 1 has a positive, strong loading value of EC, TDS, and a negative strong loading value of DO. Cluster III defines a strong positive loading value for Color and Turbidity, which are dominant in factor 1. This factor explained the total cumulative variance of 18.29%, 26.40% and 13.146% for the three clusters, respectively. The hardness of the water is associated with the rock-water interaction process in the study area. The strong loading value of TDS and EC might come from the textile effluent and the dissolution of rock and minerals in sediment.

Factor 2 explains the total variance of the three clusters accordingly: 16.01%, 20.76%, 12.54%. Cluster I show a strong positive loading value for Color and Nitrate, which dominates factor 2. For cluster II this factor shows a strong positive loading value for Fluoride, a moderate positive loading value for pH and iron, and additionally a strong negative loading value for Nitrate. Factor 2 has a strong positive loading value for Sulfate and Hardness in Cluster III. The overuse of nitrogenous fertilizer in the study area highlights the presence of Nitrate (Gutiérrez et al. 2018). The strong loading of sulfate concentration is associated with the deposition of the atmosphere and the bacterial oxidation of sulfur compounds (Sidle et al. 2000). Factor 3 accounts for 10.30%, 17.04%, and 11.52% of the total cumulative variance in the three clusters, respectively. This factor indicates that the strong positive loading of Turbidity and a moderate loading of Mn were observed in Cluster I. Table 3 shows that a strong positive loading value of Mn and the moderate loading value of Color dominate Cluster II. It has a moderately positive loading value of DO and a negative moderate loading value of Nitrate in Cluster III. The Mn content in the study area is because of the Phosphatic fertilizer plant. Dissolved and suspended materials and a brown shade in drinking water often come from rust in the water pipes, which causes Color in drinking water (Liu et al. 2017).

Factor 4 explains the cumulative variance of 9.93%, 15.99%, and 11.01% in three clusters. It is dominated by the positive, strong loading value of Fluoride in Cluster I. In Cluster II, factor 4 has a strong positive loading value for Sulfate and Ammonia. Table 3 illustrates that this factor has a strong positive loading value of Mn and a moderate negative loading value of pH in Cluster III. The negative loading values of pH indicate that the pH variations control the major ion concentration in the study area (Varol & Davraz 2014). Chemical and pharmaceutical manufacturing industries are the main sources of pollution for groundwater for Ammonia (Ma et al. 2012). Factor 5 reciprocally explains the majority of 9.43%, 11.14%, and 10.16% of the total cumulative variance. Ammonia is a strong positive loading of factor 5 both in Cluster I and Cluster III. Turbidity has a positive, strong loading value of factor 5 in Cluster II, indicating that the ion comes from the textile and dairy industries. Factor 6 explains the total cumulative variance of 9.13% and 9.61% in Clusters I and III. Thus, factor 6 has a positive, strong loading value of iron for both clusters. A moderate loading value of Sulfate in Cluster I. Iron's strong positive loading value is interlinked with sewage pollution, industrial waste, and soil erosion. (Howladar et al. 2021). In another study conducted by Rahaman et al. (2019) in Rajshahi city, Bangladesh, it was found that groundwater in the area contained high concentrations of magnesium and sulphate. These elevated levels were attributed to natural factors such as the weathering of silicate, calcareous, and clay minerals, as well as the flow through the aquifer lithology. These FA results underscore the co-occurrence of elevated iron, ammonia and nitrate concentrations within the same zones levels that exceed both national (ECR 2023) and international (WHO 2022) guideline limits. Such simultaneous chemical imbalances can lead to synergistic health effects, including gastrointestinal distress, methemoglobinemia risk in infants (from high nitrate), and long-term impacts on renal and neurological function (from excess iron and ammonia) (Ige et al. 2019; Wu et al. 2020). These findings highlight the urgent need for targeted monitoring and intervention strategies in the most affected clusters to safeguard public health.

In this context, the detection of E. coli in Gazipur City's drinking water is particularly concerning, as fecal contamination is closely linked to elevated risks of gastrointestinal disease (Hasan et al. 2019) and can precipitate diarrheal outbreaks comparable in scale to cholera (Qadri et al. 2005). Furthermore, nitrate concentrations exceeding 10 mg/L (WHO guideline value) pose a serious threat of methemoglobinemia in infants (Muhib et al. 2023), and long-term exposure has been associated with adverse reproductive outcomes, including preterm births and congenital anomalies (Lin et al. 2023). Consequently, it is crucial to implement necessary measures and policies to address bacterial contamination in drinking water at restaurants and tea stalls in Bangladesh. Given that WHO guidelines recommend zero E. coli per 100 mL of drinking water, the detection of E. coli in up to 70% of samples in some zones represents a serious public health concern and a violation of both national (ECR 2023) and international standards. Therefore, immediate intervention is warranted through stronger regulatory oversight, routine water quality monitoring, and public health education to reduce bacterial risks in drinking water served at restaurants and tea stalls in urban Bangladesh.

The study conducted in Gazipur City, a major industrial hub in Bangladesh, provides critical insights into the quality of drinking water in local restaurants and tea stalls, highlighting the broader environmental and public health impacts associated with industrial activities. The application of three distinct WQI methods (WAWQI, IWQI, and AWWQI) has revealed significant variability in water quality across different zones of the city. These findings underscore the widespread unsuitability of water for safe consumption, with water quality frequently exceeding the maximum allowable limits for various physicochemical and bacteriological parameters as defined by national (ECR) and international (WHO) guidelines.

Among the WQI methods, WAWQI and IWQI demonstrated more reliable assessments of water quality across the three identified clusters, while AWWQI was less accurate in reflecting the water quality conditions in these zones. FA further emphasized the strong associations between EC, color, Mn, ammonia, turbidity, and iron, indicating the prominent role of industrial effluents and urban pollution in contaminating the drinking water supply. To address these issues, we recommend that local authorities implement routine WQI-based monitoring and multivariate analyses to detect contamination hotspots; enforce stricter effluent discharge limits for nearby industries with real-time compliance checks; and support the installation of point-of-use treatment systems (e.g., UV disinfection, filtration) at restaurants and tea stalls.

This study also stresses the importance of incorporating multivariate analysis methods, such as WQI and FA, into routine water quality monitoring programs. These methods offer valuable insights into contamination patterns and pollutant sources, which can be used to develop more effective regulatory frameworks and intervention strategies. Furthermore, we advise conducting seasonal and longitudinal sampling campaigns to evaluate the efficacy of implemented controls and to track long-term public health outcomes.

In conclusion, effective collaboration between government agencies, local industries, and community stakeholders is essential to developing sustainable solutions for improving water quality and public health. By fostering such partnerships, it'll be possible to implement strategies such as public awareness raising campaigns, stricter effluent treatment in industries, and subsequent investigations and monitoring of groundwater quality will ensure the provision of safe drinking water and protect the broader environmental and human health outcomes in rapidly urbanizing and industrializing areas like Gazipur. The adoption of these solutions will contribute to the long-term sustainability of water resources and safeguard public health across the region.

N/A

There are no ethical issues involved in the thesis.

All authors agreed to the published version of the manuscript.

All authors have agreed to publish this article.

M. Rahadujjaman, R. Hasan, M. R. Karim and M. S. Hossain: Conceptualization, Validation, Formal Analysis. M. H. R. B. Khan, M. R. Karim, M. Rahadujjaman, R. Hasan, and A. Ahsan: Data Curation, Methodology, Visualization, Formal Analysis and Writing-Original Draft. M. H.R. B. Khan, M. R. Karim, and A. Ahsan: Writing-Review & Editing.

This research received no external funding.

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

The authors declare there is no conflict.