Sobol sensitivity analysis for non-carcinogenic health risk assessment and water quality index for Kohgiluyeh and Boyer-Ahmad Province, Western Iran

2.1 Study area

Kohgiluyeh and Boyer-Ahmad is an Iranian province with Yasuj as its capital. This province that extends 16,249 km² and is bordered by Chaharmahal and Bakhtiari province in the north, by Khuzestan province in the west, by Bushehr province in the south, and by Fars and Isfahan provinces in the east is located along the Zagros Mountains, with the geographical coordinates of 29° 36′ 00″ and 31° 26′ 00″ North latitude and 49° 55′ 00″ and 51° 53′ 00″ East longitude. Rainfall in the province has decreased from the north-east to the south-west as a result of its geographic position, where around 80 % of it is surrounded by mountains. In addition, the area's average annual rainfall is approximately 863 mm, and the average temperature is 14℃ (Fakhir Nasab et al., 2020). As indicated in Fig. 1, 130 wells were sampled from the 8 counties of Kohgiluyeh and Boyer-Ahmad.

Close

Fig. 1

Map of the study area and water sampling points.

Export to PPT

2.2 Sample collection and analysis

Local government authorities frequently evaluate the water quality of reservoirs. In Kohgiluyeh and Boyer-Ahmad province, there are eight reservoirs that were surveyed in the current research. The samples were taken from January to December 2021 and the water quality was evaluated weekly at each monitoring station. pH, Electroconductivity (EC), turbidity, Total Dissolved Solids (TDS), total alkalinity, total hardness, permanent hardness, temporary hardness, bicarbonate alkalinity, F^-, Cl^-, SO₄^2-, PO₄^2-, NO₂^–, NO₃^–, Ca²⁺, Mg²⁺, K⁺, Fe²⁺, and Mn²⁺ were some of the monitored indicators. When the highest concentrations of water quality indicators were less than the detectable level, water quality was assessed using half of the detectable limit. Standard methodologies were used for the analysis. Before sampling, the samples were gathered in disposable polyethylene containers that had been rinsed three times. A portable pH meter was used to monitor the pH levels in place. Duplicates of every sample were also taken to guarantee the validity and accuracy of the data. In fact, the accuracy of the analyses were checked using verified reference materials under identical conditions and the dependability of the data was ensured (Zhang et al., 2018). Overall, 580 samples from the entire province were gathered over the course of a year and analyzed for various parameters in the dependable lab of the water and wastewater company using the techniques from the Standard Methods (Rice et al., 2012).

2.3 Entropy water quality index

EWQI is a comprehensive water quality evaluation method that has been widely used to determine the quality of water for drinking purposes (Su et al., 2018, Zhou et al., 2016). The steps for calculating the EWQI are as follows:

The first step in calculating the EWQI is to give each water quality measure the appropriate weight. The entropy ej and entropy weight wj can then be computed using Eq. (1) to Eq. (3).

The second step in calculating the EWQI is to compute the quality rating scale (qj), which can be done using Eq. (4).

Where Cj is the concentration of the investigated water quality parameters and Sj is the water quality parameter recommendation value in the Iranian standard (ISIRI, 2010).

Finally, the EWQI can be determined using Eq. (5).

According to the findings of the EWQI, water quality can be divided into five categories, namely excellent quality (less than25), good quality (25 ∼ 50), medium quality (50 ∼ 100), poor quality (100 ∼ 150), and extremely poor quality (greater than150) (Adimalla, 2021, Su et al., 2018, Subba Rao et al., 2020).

2.4 Health risk assessment

Humans are exposed to toxic compounds in water through three major routes; i.e., ingestion through drinking water, intake through inhalation, and dermal exposure (Wang et al., 2021, Li et al., 2021, Ji et al., 2020). Since water is mostly utilized for drinking and bathing, the major exposure methods for the inhabitants in the study region were drinking water ingestion and dermal exposure. F^–, NO₃^–, NO₂^–, Mg²⁺, Mn²⁺, and Fe²⁺ were chosen to do the non-carcinogenic health risk assessment according to the measurement data. In this research, age and sex were both regarded as the factors affecting the health risk, and three groups (children, adult females, and adult males) were deemed (Zhang et al., 2021).

Eq. (6) and Eq. (7) were used to assess the non-carcinogenic risk of drinking water consumption (He et al., 2021, He and Wu, 2019).

Where IR was the ingestion rate of drinking water (L/day) and C was the contaminant concentration in the collected samples (mg/L). IR was divided into two components in this research. The first was direct water intake (IR₁), while the second was indirect water intake (IR₂) such as consuming food prepared with water. EF referred to the exposure frequency (day/year), and its value was 365 day/year. ED stood for the exposure duration (year), F (unit less) denoted the fraction of the skin's surface area in contact with water, BW referred to bodyweight (kg), AT was the average time of non-carcinogenic effects (day) (Karunanidhi et al., 2021), and RfD was the reference dosage (mg/(kg.day)). Based on the Integrated Risk Information System (IRIS) database of the USEPA, the RfD_oral levels for F^–, NO₃^–, NO₂^–, Mg²⁺, and Mn²⁺ were 0.06, 1.6, 0.1, 11.0, and 0.14 mg/(kg.day), respectively (USEPA, 1996, Council and Committee, 1977). Additionally, the provisional-RfD for Fe²⁺ was 0.7 mg/(kg. day) (Agency, 2006). The particular values for each of these have been listed in Table 1 (Yin et al., 2021).

Eq. (8) to Eq. (11) were used to estimate the non-carcinogenic risk of dermal exposure (Li et al., 2017).

Where EV stood for daily dermal contact exposure frequency that was 1 in this study, SA was the surface area of the skin (cm²), and K was the permeability of the skin that was considered 0.001 cm/h. The contact time (h/day) or exposure time (ET) was set at 0.4 h/day for both adults and children (Li et al., 2016). ABS_gi was a gastrointestinal absorption factor with a value of 1 for all the ions and CF was a conversion factor with a value of 0.001. Eq. (12) and Eq. (13) were used to calculate the total non-carcinogenic risks.

Where HI referred to the non-carcinogenic risk index. HI less than 1 denoted a tolerable risk, whereas HI ≥ 1 indicated a higher-than-tolerable non-carcinogenic risk (Zhou et al., 2021).

2.5 Sobol sensitivity analysis

To find the relevant input parameters and evaluate their impacts on the variance of the exposure outcomes, use was made of Sobol sensitivity analysis combined with the Monte Carlo approach (Sobol, 2001). The Sobol Sensitivity Indices (SIs) refer to the ratio of the partial variable to the overall variable. The first term of SI is named First Order Sensitivity Index (FOSI), which explains the impact of a single variable on the variance in model results. The second term is named Second-Order Sensitivity Index (SOSI) that explains the impact of the interaction between variables. Finally, the Overall Order Sensitivity Index (TOSI) is used to compute the overall influence of the variable on the final variance. Sensitivity indices greater than 0.1 (highly sensitive), 0.01–0.1 (sensitive), and less than 0.01 (insensitive) represent significantly relevant, prominent, and unresponsive input variables, respectively (Mukherjee and Singh, 2022, Mukherjee and Singh, 2021). In the current research, Sobol sensitivity analysis was carried out utilizing the R-platform (‘EnvStats’, ‘sensobol’, ‘EnviroPRA’ packages), version 4.1.2.

2.6 Spatial distribution

This investigation revealed the spatial distribution of the health risk assessment. The Arc-GIS program (10.8.1 version) was utilized to analyze the spatiotemporal variations in the concentrations of contaminants in groundwater in this area to detect high contamination locations by Inverse Distance Weighted (IDW) analysis (RadFard et al., 2019).

3.1 General hydrochemistry

Based on the findings of laboratory and field experiments, the chemical features of groundwater were detailed. The statistical information on the appropriateness of groundwater for drinking based on Iranian Standard (ISIRI, 2010) and World Health Organization’s (WHO) guidelines (Organization, 2017) has been presented in Table 2. Based on the results, the mean temperature of groundwater over a year was 19.44 °C, which contributed to regulating the responses in the groundwater and aquifers media. In addition, the groundwater in the region was considerably mineralized. The average EC and TDS were 567.4 µmhos/cm and 319.1 mg/L, respectively, indicating that the liquid phase of water had a considerable contact with both gaseous and solid media. None of the samples exceed the Iranian Standard and WHO-recommended limits for EC and TDS (Organization, 2017, ISIRI, 2010). Moreover, the mean pH of the groundwater was 7.64, which indicated that it was moderately alkaline. As seen by the mean concentration of 195.2 mg/L in groundwater, the total alkalinity of groundwater was mostly attributable to soluble carbon dioxide molecules in water. Furthermore, the average hardness of the groundwater in the area was 239.3 mg/L as CaCO₃, indicating that the groundwater in the region was extremely hard. Water with a hardness of more than 200 mg/L in 58.1 % of the samples was unfit for human consumption (Organization, 2017).

The mean Ca²⁺ and Mg²⁺ levels in groundwater were 58.65 mg/L and 20.01 mg/L, respectively. Sandstone, limestone, basalts, and dolomites found in the region might considerably contribute to Ca²⁺ and Mg²⁺ levels in groundwater. In all WHO-mandated samples, the quantity of Ca²⁺ ions were within the allowed level, but the level of Mg²⁺ ions surpassed the allowed limit in 3 % of the samples. Additionally, the mean K⁺ content was 2.42 mg/L, which might be due to the potential for K⁺ to be absorbed back through groundwater movement. K⁺ is produced by the ion exchange process and can come from both geological and human origins in groundwater. If potassium compounds are widely utilized, farming can add K⁺ to groundwater. Moreover, the concentration of Mn²⁺ in groundwater was averagely 0.17 mg/L. According to the WHO guidelines, the allowable level of Mn²⁺ is 0.1 mg/L. The WHO’s guideline was exceeded in almost 45 % of the samples. Finally, the Fe²⁺ concentration differed widely from well to well, with a mean of 0.18 mg/L. The WHO has established a maximum of 0.3 mg/L for Fe²⁺ in drinking water (Nawale et al., 2021).

In the research region, SO₄^2– levels in groundwater varied from 1 to 612 mg/L (mean: 55.48 mg/L). In 6 % of the samples, the level of SO₄^2– ions surpassed the allowed limit. Groundwater may contain SO₄^2– as a result of atmospheric deposition. It is impossible to rule out the potential of sulphur-bearing minerals like pyrite occurring in coal. Sulfate fertilizers can play a role, as well (Marghade et al., 2021, Organization, 2017). On the other hand, phosphate concentrations were low (0.24 mg/L). The highest phosphate level was 1.1 mg/L, while the lowest value was 0.01 mg/L. The mean NO₂^– and NO₃^– concentrations in groundwater were 0.06 and 13.41 mg/L, respectively. Inappropriate disposal, sewage, and farming in the region were the substantial contributors to NO₃^– in groundwater. The NO₃^–exceeded the Iranian standard and WHO’s limit of 50 mg/L in 1 % of the samples. Finally, the level of F^– varied from 0 to 1.3 mg/L (mean: 0.45 mg/L). Aqueous media with F^– carrying minerals caused F^– to enter the groundwater in the studied region (Sankhla and Kumar, 2018, Organization, 2017, ISIRI, 2010).

3.2 Water quality assessment

F^-, NO₃^–, NO₂^–, Ca²⁺, Mg²⁺, Mn²⁺, and Fe²⁺ were chosen as water quality parameters in this research to accomplish an all-water quality evaluation utilizing the EWQI. The EWQI calculations for the groundwater samples have been provided in Table 3. Accordingly, EWQI was 80 to 516 for the measured variables in a year. One reservoir had medium-quality water (rank 3), while the other had severely poor-quality water (rank 5) that could be utilized for drinking after considerable treatment. Overall, the findings demonstrated that the quality of groundwater was typically unfit for drinking throughout the year. The evaporating and concentrating impacts of the groundwater might be attributed to high temperatures and high evaporation rates throughout the year. Furthermore, the increase in surface water and groundwater levels over a year made it simpler for surface water, which could be contaminated by animal feces, human excrement, and other sources, to permeate into groundwater.

Drinking water is transmitted by pipelines and increased temperature will alter the development of microorganisms in pipelines, resulting in a rise in bacteria, microorganisms, and byproducts and lowering the overall quality of water. Mg²⁺ and NO₃^– are two important ions for determining water quality. Agricultural activities are linked to NO₃^– concentrations in various seasons. The water quality of the groundwater is often inadequate for house use. The water quality of Gachsaran was better than that of other reservoirs, which suggested that climate change could have a big influence on the groundwater quality (Ji et al., 2020).

3.3 Health risk assessment

Since groundwater is the primary supply of drinking water, the health risks associated with ingesting polluted water gets to be a major concern (Adimalla, 2021). The Hazard Quotient (HQ_oral) for the consumption of drinking water in adult males ranged from 0.955 to 1.492, with an average of 1.179. This measure ranged from 0.877 to 1.371 in adult females, with an average of 1.083. For children, the HQ_oral through the consumption of drinking water varied from 1.061 to 1.658, with an average of 1.310. Based on the results, 75 % of the samples for adult males, 50 % for adult females, and 100 % for children had an HQ_oral surpassing 1, indicating that almost all the samples might cause non-carcinogenic risks. Dermal exposure had a lower hazard quotient (HQ_dermal) compared to oral exposure, ranging from 0.003 to 0.004, 0.002 to 0.004, and 0.004 to 0.007 for adult males, adult females, and children, respectively. This showed that the non-carcinogenic risk was primarily transmitted through the oral ingestion of polluted water. The overall non-carcinogenic health risks (HI_total) for children, adult females, and adult males have been presented in Table 4. Accordingly, adult males had a non-carcinogenic risk of 0.958 to 1.497 (mean: 1.182), adult females had a non-carcinogenic risk of 0.879 to 1.374 (mean: 1.086), and children had a non-carcinogenic risk of 1.065 to 1.664 (mean: 1.315). These findings suggested that adults and children in the study region encountered a significant non-carcinogenic health risk. On the other hand, children in the research region faced higher non-carcinogenic health risks in comparison with adults. However, no significant difference was observed between the adult males and females in terms of health risks. Similar studies have also revealed a greater health risk for children compared to adults, implying that children were more sensitive to pollutants compared with adults (Li et al., 2021, Wang et al., 2021). The non-carcinogenic health risks associated with the Bahmai reservoir were greater than those associated with other reservoirs.

Pollutants lead to health concerns in diverse ways. As indicated in Fig. 2, Mg²⁺ (57.93 %) followed by NO₃^– (39.80 %) were the most important factors in non-carcinogenic risks. Other pollutants contributed by less than 5 % to non-carcinogenic risks, implying that Mg²⁺ and NO₃^– might be the causes of negative human health impacts. The sequence, in which the contaminants contributed to non-carcinogenic risks was as follows: Mg²⁺ > NO₃^– > F^- > Mn²⁺ > Fe²⁺ > NO₂^–.

Close

Fig. 2

Contributive ratios of non-carcinogenic risk pollutants to health risks.

Export to PPT

Generally, there are three uncertainties in every health risk assessment: model uncertainty, parameter uncertainty, and input data uncertainty. In the present study, there were some uncertainties in the model itself. For instance, SA and DA were computed using experimental models and, consequently, the SA and DA values varied depending on the used experimental models. In addition, assessment models contained a large number of factors, each of which was correlated to a level of uncertainty. For adults and children, the water consumption rate was considered 0.52 to 1.73 L/d in a direct way and 0.21 to 0.57 L/d in an indirect way in this research. These estimates were based on statistical data, but they varied from person to person. Various rates of consumption result in different dangers. Likewise, BW, AT, and EV quantities are all derived from statistical data, and various people may have various parameter quantities. Furthermore, the input data had an element of uncertainty. Even though all the water samples were evaluated in a qualified laboratory using conventional processes and techniques, both analytical and systematic errors were unavoidable and added to the level of uncertainty. Despite these concerns, the study findings were credible and useful. The findings indicated the mean risk levels, with individual risks varying within a reasonable range around the average. These results are particularly important for decision-makers, since they provide them with broad risk information that leads to making informed decisions based on the majority of people. When comprehensive information is supplied, it is often difficult to make sound decisions. Decision-makers, for instance, find it hard to make judgments if they take into account each individual's real-life circumstances (Li et al., 2016, Qasemi et al., 2019, Nabizadehb et al., 2019).

The drinking water supplies of several towns, including Tehran, Mashhad (Khorasan Razavi), Ardabil, Ahwaz (Khuzestan), and Qom, have greater nitrate concentrations than the level advised by the Institute of Standards and Industrial Research of Iran (ISIRI) and WHO. This is according to a study by Darvishmotevalli et al. in Iran, which evaluated the carcinogenic hazards associated with nitrate exposure in drinking water. As a result, the cancer risk associated with nitrosamine, which is created when nitrate is taken through drinking water, is higher in these cities. According to this study, there is a significant link between nitrate exposure and the occurrence of cancer in the vast majority of Iranian districts whose drinking water sources have been restricted (Darvishmotevalli et al., 2019). The level of nitrate in drinking water was identified by Fallahzadeh et al. as the most crucial factor. The average nitrate concentration was 27.57 to 6.80 mg/L, and all observed amounts were below the allowed maximum threshold. Children in this group were at a high risk since their HQ values were more than 1. The author came to the conclusion that lowering the concentration of nitrate might lower population risk (Fallahzadeh et al., 2019).

3.4 Sobol sensitive analysis

The Sobol sensitivity analysis was performed to analyze the most relevant input factors related to the non-carcinogenic health hazards related to nitrate contamination in drinking water in the population of the region. The factors for this model have been listed in Table 1. As the table depicts, the oral model had four input factors, namely nitrate concentration in groundwater (C_w), Body Weight (BW), Exposure Frequency (EF), and Intake Rate (IR₁ and IR₂), while the dermal model had six input factors as follows: contaminant concentration (C_w), the fraction of skin contact (F), exposure during shower time (ET), BW, skin surface area (SA), and EF.

According to the results of Sobol sensitivity analysis (Fig. 3), the C_w parameter was the most effective parameter in oral HQ among the children (0.793), males (0.386), and females (0.466). Following C_w, the parameters were ranked as IR (0.093) > EF (0.024) > BW (0.007) in the children and as IR (0.321, 0.274) > BW (0.022, 0.022) > EF (0.011, 0.014) in male and female adults. The BW value was less than 0.01 among the children, indicating its insensitivity in terms of relevance. Besides, the EF value was the less sensitive parameter in adult males and females. However, since the C_w value was more sensitive in females than in males, it could be concluded that this parameter was more important in females, while IR was more important in males. Since the total sensitivity was higher than the first-order sensitivity, these parameters had interactions. The values of the effective parameters in total Sobol sensitivity in oral HQ were as follows: C_w = 0.874, IR₁ = 0.156, EF = 0.042, and BW = 0.013 for children, C_w = 0.627, IR₁ = 0.559, BW = 0.055, and EF = 0.029 for males, and C_w = 0.675, IR₁ = 0.474, BW = 0.051, and EF = 0.032 for females. The interaction diagram (Fig. 3) for the children showed that C_w had interaction effects with three factors, with the effect being significantly stronger for C_w-IR₁ compared to C_w-EF and C_w-BW, which varied for different IR₁ values. There were also interactions between C_w-BW, C_w-IR₁, BW-IR₁, C_w-EF, and IR₁-EF in both males and females. However, the severity of this interaction was much greater between IR₁ and C_w, and all the other parameters, except for EF-BW, had very little interactions. Furthermore, the children were more exposed to non-carcinogenic risks from water contaminants in comparison with adults, and C_w was among the most effective parameters in the oral HQ of this group. The sensitivity study findings for IR₂ were comparable to the sensitivity analysis results for IR1, which was due to the same parameters and the same way of exposure to nitrate in water (Fig. 4). The effective parameters in oral HQ in total Sobol sensitivity were C_w = 0.793, IR₂ = 0.093, EF = 0.024, and BW = 0.007 for children, C_w = 0.386, IR₂ = 0.321, BW = 0.022, and EF = 0.011 for males, and C_w = 0.466, IR₂ = 0.274, BW = 0.022, and EF = 0.014 for females.

Close

Fig. 3

Sensitivity analysis based on the oral HQ ingestion model (C_w, BW, IR₁, EF) for the three age groups considering the first-order effect (S_i), total effect (T_i), and pair-wise interactions.

Export to PPT

Close

Fig. 4

Sensitivity analysis based on the oral HQ ingestion model (C_w, BW, IR₂, EF) for the three age groups considering the first-order effect (S_i), total effect (T_i), and pair-wise interactions.

Export to PPT

C_w was the most important and effective parameter in dermal HQ amongst children (0.837), males (0.783), and females (0.792). The parameters were ranked as F (0.064) > EF (0.025) > BW (0.008) in the children and as F (0.060, 0.060) > BW (0.044, 0.038) > EF (0.024, 0.024) in male and female adults. All the ET and SA values were less than 0.01, showing their insensitivity to the dermal HQ. Furthermore, because the C_w value was larger in females than in males, this parameter was more essential in females. However, BW was more important in males. Moreover, because the total sensitivity was greater than the first-order sensitivity, it might be inferred that these factors interacted with each other. Considering the total Sobol sensitivity in dermal HQ, the effective parameters were C_w = 0.901, F = 0.108, EF = 0.044, and BW = 0.013 for children, C_w = 0.868, F = 0.104, BW = 0.077, and EF = 0.042 for males, and C_w = 0.874, F = 0.105, BW = 0.067, and EF = 0.043 for females. In the children, the interaction diagram (Fig. 5) revealed that C_w interacted with three parameters, with the influence of C_w-F being much greater than C_w-EF and C_w-BW, which changed for various F values. In both males and females, there were interactions between C_w-F, C_w-BW, and C_w-EF, with the intensity of the interaction between F and C_w being significantly stronger compared to the other parameters. All the other parameters had moderate interactions. Similar results were also obtained in another study conducted by Karunanidhi et al. on nitrate concentrations amongst children, males, and females. The results also indicated that children were more sensitive to changes in the parameters affecting health risk compared to males and females (Karunanidhi et al., 2021).

Close

Fig. 5

Sensitivity analysis based on the dermal HQ model (C_w, BW, SA, F, ET, EF) for the three age groups considering the first-order effect (S_i), total effect (T_i), and pair-wise interactions.

Export to PPT

3.5 Spatial distribution

F^-, NO₃^–, NO₂^–, Ca²⁺, Mg²⁺, Mn²⁺, and Fe²⁺ all showed a wide range of contamination from mild to severe. Nonetheless, the HI of the three groups (children, males, and females) revealed the comparable geographical distribution patterns of the environmental risk (Fig. 6a, b, and c). The hazard in this area was substantially higher compared to other places. Additionally, the concentrations of these ions were higher in the west, northwestern, and bordered areas with Fars province than in other regions. This provided obvious environmental consequences inside these zones, needing more monitoring. Given that groundwater is the principal supply of water in Kohgiluyeh and Boyer-Ahmad province, it is logical to assume that the ions concentrations are rather higher in these resources than in other regions (Davraz and Batur, 2021). Ion concentrations have risen as a result of geological composition, land use in the vicinity of groundwater sources, pollution leakage from diverse sources, and industrial effluent infiltration. Furthermore, ions concentrations alter with precipitation and annual rainfall, and groundwater levels shift from season to season. In places with high ion concentrations, the watershed height and depth are usually low. Another consideration is the general east-to-northwest slope of Kohgiluyeh and Boyer-Ahmad plains. As a result, agricultural operations should be established carefully (Ehya and Mosleh, 2018). Groundwater overuse to suit municipal, industrial, and agricultural demands can change the properties and natural components of water. Development and upgrade of the wastewater collection system as well as the rate at which industrial effluent is generated are the other factors that influence the concentrations of water pollutants (e.g., nitrate and fluoride). Monitoring of wells, reservoirs, and nearby regions is crucial due to the natural and human factors that affect the concentrations of ions in groundwater (Adeli Moghadam et al., 2020, Das and Nag, 2022).

Close

Fig. 6

Spatial distribution of HI in children (a), adult males (b), and adult females (c).

Export to PPT

Close

Fig. 6

Spatial distribution of HI in children (a), adult males (b), and adult females (c).

Export to PPT