Results of the hydrochemical analysis along with descriptive statistics of quality parameters of the groundwater samples which were taken from the studied area are presented in Table 1.
EC values in the investigated area ranged from 110 to 1750 μs/cm, these values are much less than the values reported by Baghvand et al. in an aquifer in Iran central desert (1987–12751 μs/cm) . Figure 2 represents EC variation in Tabriz, indicating high concentration of salts in east and a zone in north of the studied area. In terms of hardness, water is grouped as soft water (>75 mg/L CaCO3), medium hard (75–150 mg/L CaCO3), hard water (150–300 mg/L CaCO3) and very hard water (>300 mg/L CaCO3). Range and average values of total hardness in the sampled water were 52 to 476 and 185.88 ± 106.56 mg/L as CaCO3, respectively, representing that the studied water samples could be grouped as hard water (Figure 3). In a similar study in India, majority of the samples fall in very hard water category (>300 mg/L CaCO3) . Nitrate values, as an important parameter regarding its health effects, varied from 0.4 to 59.4 mg/L with average value of 11.91 ± 10.49. In the similar study in India, Nitrate concentration in 73.68% of samples exceeded the guideline value (50 mg/L) . Nitrate distribution is given in Figure 4, according to which only, in village 2 (Eskandar), level of nitrate exceeded guideline value (50 mg/L). Figures 5, 6, 7 show pH, bicarbonate and fluoride variations in the studied area, respectively. As can be seen, in most parts of the studied area, pH values were above 7; however, in some parts like north, east and southwest, this value was below 7. In many parts of the considered area, bicarbonate concentration was below 250 mg/L as CaCO3. For fluoride, concentration was less than 0.5 mg/L, except in villages 3 and 28.
Tables 2 and 3 demonstrate percentage of constituents at the sampled points and hydrochemical faces of the groundwater according to piper diagram, respectively. According to these tables, in most parts of the considered area, hydrochemical faces of the groundwater were of calcium-bicarbonate type. Piper diagram of the studied area is given in Figure 8. According to the diagram, groundwater was divided to six faces [4, 12, 21]. In the present investigation, bicarbonate was the most dominant anion (in 93% of the samples) and calcium was the most dominant cation (in 65% of the samples). Therefore, chemical characteristic of the water was dominated by Ca HCO3 type water.
Table 4 shows concentration of heavy metals and trace elements in the groundwater samples. In addition to the heavy metals presented in Table 4, some other elements including cadmium, cobalt, iron, manganese, molybdenum, nickel, lead, antimony, selenium and vanadium, were either analyzed which their concentrations were non-detectable in samples.
One of the heavy metals of concern in the studied area was arsenic, which is a naturally occurring contaminant in groundwater. Exposure to high levels of arsenic can cause short-term or acute symptoms as well as long-term or chronic health effects. Arsenic in groundwater is found largely due to the minerals associated with previous volcanic activities dissolving from weathered rocks, ash and soils [22–24]. However, arsenic concentrations of greater than 10 μg/L (guideline value of World Health Organization for arsenic in drinking water ) were detected in the water supplies, especially in western district of the considered area (Figure 9). In Shadabad, Kondrud, Asbes, Varanag, Esfahlan and Khellejan villages, concentration of above 10 μg/L was found.
The presence of aluminum at concentrations exceeding 0.1–0.2 mg/l often leads to consumer complaints due to its deposition of aluminum hydroxide floc. There is no health-based guideline value for the concentration of aluminum in drinking water; however, aluminum concentrations of less than 0.1 mg/L are achievable in many circumstances . Results of this investigation demonstrated that only 2 of the samples had aluminum concentrations of more than 0.1 mg/L.
Inorganic mercury is a predominate form that is found in surface and groundwater, usually at concentrations of below 0.5 μg/L; however, local mineral deposits may produce higher levels in groundwater . As depicted in Table 4, only in one case, mercury concentration was above the guideline value (6 μg/L). Main problems caused by inorganic mercuric poisoning include liver and renal damage which lead to death . Organicmethylmercury affects the central nervous system. Relatively few instances of elevated concentrations of mercury in groundwater have been reported, except when mercury is included among other contaminants in site-specific hazardous waste investigations or other studies of groundwater contamination in industrialized areas [26, 27].
Other heavy metal values presented in Table 4 were below the standard values.
Table 5 provides a matrix of correlation coefficients between quality parameters of the analyzed water samples along with heavy metals. Values of high correlation are specified in bold. As can be observed in the table, there was high correlation between total hardness, Ca2+, HCO3−, Mg2+, K+, Na+, SO42−, Cl−, B and Sn. Furthermore, there was significant correlation between Na+, SO42−, NO32− and B. The correlation between Ca2+, NO32− and Hg was also significant. As can be observed in Table 5, the correlation between Cr and Br was negatively significant.
Table 6 demonstrates rotated factor loadings for water quality parameters. In KMO and Bartlett’s Test, p > 0.001 with coefficient of 0.658 was significant. PCA showed that 3 components could cover 84.3% of the parameters. In 32 analyzed water samples, according to rotated component matrix with 3 factor solution, PC1 accounted for more than 60.3% of total variance in the dataset and was loaded with magnesium, potassium, bicarbonate, sulfate, hardness and electric conductivity; i.e. these parameters demonstrated a similar behavior in groundwater. The second component (PC2) explaining 14.8% of total variance had strong positive loadings for nitrate, calcium, chloride and sodium. The third component (PC3) of PCA demonstrated that only 9.1% of total variation had positive loading of pH and fluoride.
In KMO and Bartlett’s test for heavy metals, p > 0.001 with coefficient of 0.535 was considered significant. Principal component analysis showed that three components explained 72.39 percent of variance (Table 7). Using rotated component matrix with 3 factor solution, PC1 included boron, tin and mercury. PC2 contained copper, zinc, barium, chromium and beryllium and PC3 was loaded with arsenic and aluminum. In PC2, zinc, barium and chromium had reverse correlation with beryllium and copper and also there was reverse correlation between arsenic and aluminum in PC3.Figure 10 demonstrated the dendrogram obtained by CA for water quality parameters. The figure indicates relationship and similarity between water resources. This dendrogram introduced four distinct groups as A, B, C and D. Considering the location of different villages in the already presented quality maps, it can be observed that the villages grouped in cluster A (3, 5, 7,14, 15,19, 20, 22, 23, 25, 26, 27, 28 and 29) were mainly located in southern and northern parts of the studied area, the ones grouped in cluster B (1, 4, 10, 13, 18, 21, 24, 30 and 31) were distributed in western part of this area and those located in cluster C (6, 8, 9, 11, 12, 16, 17 and 32) were seen in different parts. Finally, the dendrogram clarified abnormality of the water sample from Eskandar village which constituted one group as cluster. As shown in Figure 10 and the results presented in tables and maps thus far, quality of water in this village was infelicitous compared to other villages, which could be due to the fact that water well of this village was located close to the river and in the agricultural area. Since water table level in this area had high intrusion of contaminants through river, which passed through the village and also agricultural drainage to this water table could be the probable reason for this difference.
Application of CA is useful for classifying groundwater in the whole region and makes adequately serving for spatial assessment possible in an optimal manner. Therefore, the number of sampling sites and cost in the monitoring network is reduced without losing any significance of the outcome [28, 29].