Evaluation of Geochemical Controlling Parameters of Groundwater in Arak using Factor Analysis

By Feridon Ghadimi Arousmahaleh¹ * and Mohammad Ghomi²
February 2012

The authors are from the Department of Mining Engineering, Arak University of Technology, Arak, Markazi, Iran

  1. Assistant Professor *Corresponding Author
  2. Head, Mineral Labs
Abstract
In this Study, the quality of groundwater examined relative to WHO standard and identified the controlling factors quality of groundwater using multivariate statistical tool such as factor analysis. This study finds that the major water type in the aquifers of Arak is the Ca-HCO3 and Ca-HCO3-Cl water type, which gradually degrades into very low saline Ca-Na-HCO3-Cl water types towards the Mighan playa. Factor analysis identified three factors that explained 76.24% of the total variance. The first factor shows the relationship between HCO3, Fe, TDS concentrations and is factor of dissolution of carbonate rocks such as limestone formation or natural factor. The second factor shows relationship between Na, Cl, SO4, Ca and NO3 concentration and is intrusion playawater, anthropogenic factor and in the third factor, F, pH shows natural factor. Geographical distribution of scores of factors identifies areas with higher second factor, concentrated in northeast of area and towards Mighan playa.

Keywords: Groundwater, Factor Analysis, Quality, Aquifer, Arak

1. Introduction

The quality of groundwater is controlled by several factors, including climate, soil characteristics,  manner of groundwater through the rock types, topography of the area, saline water intrusion in playa and coastal areas, human activities on the ground, etc (Das Brijraj and Kaur, 2007; Cloutier et al., 2008). As the study area consists of present sediment in the center and rocks around of study area, the groundwater potential is governed by several factors like weathering, anthropogenic impact and saline water intrusion. Groundwater samples were collected and analysed for major ions. To compare the different chemical parameters of the water analyses and determine the relationship between them, factor analyses were applied to a large number of water quality data samples. A form of principal component analysis is used, transforming the set of variables into factores, which, successively, extract a maximal part of the variance in the data set (Reghunath et al., 2002; Love et al., 2004; Prasanna et al., 2010).

In groundwater quality management, it is important to relate the spatial distribution of different chemical parameters to different possible sources. Different sources have characteristically different chemical signatures, for example, a sewage work is liable to be associated with high levels of nitrate. However, such theoretically probable signatures do not always occur, nor are they necessarily statistically valid. Factor analysis provides a way of identifying chemical parameters these are the statistically significant factors determined by the analysis. Contouring the factor scores onto a map allows the distribution of significant factors to be compared spatially to the location of possible sources.

Figure 1 · Map of the Study Area showing Location and Sample Locations
Figure 1

In this paper, one case studied on application of factor analysis to groundwater quality problems in center of Iran are presented. In the study area, factor analysis separates different probable sources of contamination to the local groundwater.

2. Study Area

The area of study is located in the center of Iran characterized by a semi-arid climate and an average precipitation and temperature of about 280 mm/year and 11 °C respectively (Zamani, 1999) Most of its inhabitants are concentrated in city of Arak  with more than 400000 inhabitants working mainly in the industrial plants.

The aquifer of Arak (Fig. 1) is developed into the medium to fine phases of the Pleistocene sediments, which occupy a broad graben between Mounts Arak and Ashtian. The bedrock of these formations is composed of crystalline limestones of the zone of low metamorphism rocks (Fig. 2). The studied area is situated in the alluvial plain and aquifer is directly fed by stream water coming from different reliefs surrounding the depression inter-mountainous of Mighan playa. The plain hosts a large number of water-wells with depths varying from 70 to 150 m. Most of these wells supply water for drinking. The direction of groundwater flow around Arak plain is from southwest to northeast and toward saline Mighan playa (Fig. 2).

Figure 2 · Geological map of study area and Arak alluvial plain is located in southwest of saline Mighan playa
Figure 2

3. Materials and Methods

3.1 Groundwater Sampling

Water samples were collected from shallow wells for urban water supply (see Fig.1) using standard sampling procedures during sampling campaigns in 2011. The shallow wells were drilled to depths between 70 and 150 m. All samples were analysed for main chemical descriptors using standard methods. pH (ISO, 1994) and electrical conductivity (EC) (ISO, 1985) were measured in field ,whilst the rest of the parameters (Na, Ca, Mg, Fe, HCO3, Cl, F, NO3, SO4) and total dissolved solids (TDS) were determined at the Chemistry Department laboratory, Water Organization of Markazi  Province (Table 1).

Table 1 · Chemical Composition of Groundwater (in mg/l) except EC (µs/cm) and pH
Sample pH EC Ca Mg Na Fe Cl SO4 NO3 F HCO3 TDS
1 7.44 543 81.7 15.6 50 0.04 40 70 25 0.6 180 268
2 7.6 354 59.3 9.8 36 0.03 20 80 15 0.43 160 188
3 7.29 1139 85.8 21 80 0.05 80 130 70 0.6 260 302
4 7.25 702 9.6 26.8 15 0.13 35 76 60 0.65 320 350
5 7.38 686 97 22 30 0.05 40 51 30 0.51 220 340
6 7.8 622 64 24.4 20 0.06 50 65 30 0.46 220 260
7 7.25 843 104 28.8 15 0.05 30 96 45 0.5 360 380
8 7.57 610 70.5 15.6 57 0.03 54 100 27 0.7 180 240
9 7.3 866 93 35.2 63 0.01 105 97 80 0.46 240 376
10 7.34 1064 78 25.5 65 0.07 60 77 70 0.5 220 302
11 7.36 874 109 28 80 0.1 110 274 90 0.7 250 390
12 7.3 923 97.5 25.5 43 0.06 70 90 100 0.51 250 351
13 7.35 1145 141 16.6 60 0.06 120 135 90 0.52 260 420
14 7.42 840 82.6 29.7 36 0.05 80 54 30 0.53 240 331
15 7.47 683 70.2 18.6 20 0.05 47 54 25 0.55 256 253
16 7.48 603 65.5 28 27 0.05 40 56 25 0.52 220 281
17 7.44 835 85.8 21.8 45 0.04 70 83 35 0.83 220 304
18 7.33 662 86.5 23 46 0.05 40 120 45 0.65 300 310
19 7.49 675 113 20.5 27 0.01 110 104 27 0.67 304 366
20 7.89 615 57.7 35.1 68 0.03 64 50 25 0.7 210 288
21 7.68 506 53 14.6 65 0.03 47 130 21 0.52 180 200
22 7.74 535 52.8 21.5 25 0.01 40 72 27 0.61 184 200
23 7.69 501 56 14.6 46 0.01 43 80 22 0.6 174 200
24 7.69 523 51.5 15.6 30 0.01 38 72 19 0.65 180 216
25 7.8 370 44 12.2 12 0.01 10 20 10 0.41 144 160
26 7.52 428 43.2 11.5 13 0.01 10 15 9 0.32 174 156
27 7.08 395 44.8 10.5 13 0.01 10 15 11 0.32 180 156
28 7.24 439 48 28.8 12 0.01 12 10 12.5 0.28 180 190
WHO Standard for EC=250 µs/cm, Na=200 mg/l, Cl=250 mg/l, F=1.5 mg/l, SO4=500 mg/l, NO3=50 mg/l, EU Standard for Fe=0.2 mg/l

The total concentrations of Na, Ca, Mg and Fe were determined using flame atomic absorption spectroscopy. Titrimetric methods were used to determine the concentration of Cl and HCO3. Molecular absorption spectroscopy was used to determine the amount of NO3 and SO4 (Yong et al., 1998). The amount of total dissolved solids (TDS) was determined gravimetrically by evaporation. The analytical precision for the measurements of cations and anions was determined by calculating the ionic balance error that varies by about 4-10 % (Freeze and Cherrry, 1979).

3.2 Multivariate Statistical Analysis

A regional hydro-geochemical study is intrinsically a multivariate problem because of a large number of chemical parameters (variables) associated with a large number of sampling sites (observations). They have been used extensively in the assessment of hydro-geochemical data, and have been established as useful tools in studying hydro-geochemical patterns of watersheds and significantly help to classify groundwater and to discover major mechanisms influencing groundwater chemistry. Examples of the successful use of multivariate statistical methods in hydro-geochemical studies are contained in Helstrup et al., 2007; Yidana et al., 2008; Cloutier et al., 2008; Monjerezi et al., 2011. In this study, 12 chemical variables (pH, EC, Ca, Mg, Na, Fe, Cl, SO4, NO3, F, HCO3, TDS), in 28 samples were analyzed using Factor Analysis (FA). The FA was used as a quantitative and independent approach for grouping of the groundwater samples and making of correlations between chemical parameters and groundwater samples, respectively.

Table 2 · Descriptive Statistics for the 28 Groundwater Samples from the Study Area
Parameters Mean Median Minimum Maximum Variance Std.Dev. Skewness Kurtosis
pH 7.47 7.44 7.08 7.89 0.04 0.20 0.37 -0.56
EC 678 642 354 1145 49368 222 0.59 -0.32
Ca 73 70 9.60 141 743 27 0.22 0.56
Mg 21 21 9.80 35 51 7.14 0.14 -0.80
Na 39 36 12 80 463 21 0.38 -1.03
Fe 0.04 0.04 0.01 0.13 0.00 0.02 1.21 2.19
Cl 52 45 10 120 974 31 0.66 -0.18
SO4 81 76 10 274 2590 50 1.91 6.86
NO3 38 27 9 100 715 26 1.08 -0.02
F 0.54 0.52 0.28 0.83 0.02 0.12 -0.15 0.04
HCO3 223 220 144 360 2753 52 0.81 0.32
TDS 277 284 156 420 6100 78 -0.01 -1.13

The data for most chemical parameters (in mg/l) are positively skewed (Table 2), thus all chemical parameters were standardized prior to the multivariate analyses by subtracting the mean value and dividing by the standard deviation of the parameter (Liu et al., 2003). In this study FA was performed using Statistica Software Version 7. The FA was performed using a combination of Ward’s linkage method (Ward, 1963), adopting the Euclidean distances as a measure of dissimilarity. In addition to FA; Pearson correlations were used to find inter-relationships of the chemical parameters for samples.

4. Result and Discussion

4.1 Water Drinking Quality

It is considered that results from the dry season are representative of the water quality in the study area. The results of chemical analyses and field measurements for the groundwater samples for the dry season are summarized in Table 1. The pH values from 7.08 to 7.89 were found, with an average of 7.47, indicating neutral water.  Almost all the water samples in the present study fall above the current standard for Conductivity (EC), which is 250 µs/cm (WHO, 2004) (Fig. 3a). EC varied considerably from 354 to 1145 µs/cm (mean 678 µs/cm) and correspondingly TDS ranged from 156 to 420 mg/l. All groundwater samples would be classified as fresh (TDS<1000 mg/l; Fetter, 1990). In comparison with the WHO standard (WHO, 2004) and EU standard (Comber et al., 2008) that specifies that quality of water from wells, the guidelines for Fe, Na, SO4, and F fall under WHO standard guideline (Fig. 3b, c, d, e). It is observed that 7 of the 28 sampling wells under investigation, which is about 25 %, contain NO3 above the permissible level (Fig. 3f). Therefore, the majority of the groundwater samples are suitable for drinking purposes.

Figure 3 · Concentration of hydrochemical parameters relative to WHO standard for 28 samples of water
Figure 3

In general, the electric conductivity (EC) content in groundwater increases as it flows towards the Mighan playa in northeast of study area (Fig. 4a). The high salinity along the northeast area indicates that the Mighan playa is significantly influent to the underlying water table, suggesting a slow movement of water from the playa into the aquifer (Zamani, 1999). In water samples with high total ion concentration, representing high mineralization, the dominant ions are Na, K, SO4 and Cl (Table 1). Geographical distribution of NO3 shows area with higher concentrated in northeast of study area and it reveal origin of agricultural fertilizers and human activities in general (Fig. 4b).

Figure 4 · Contours of equal EC (4a), Contours of equal NO3 concentration (4b), showing that the highest values in north-east of area
Figure 4

4.2 Hydrochemical Features

Hydrochemical data of wells were mapped on Piper Diagram (Fig. 5). The central diamond-shaped figure displays that most well waters are concentrated in three groups: (1) Ca-HCO3 type waters (samples 4, 7, 16, 25, 26, 27 and 28) :(2) Ca-HCO3-Cl type waters (samples 5, 6, 12, 14, 15 and 22): (3) Ca-Na-HCO3-Cl type waters (samples 1, 2, 3, 8, 9, 10, 11, 13, 17, 18, 19, 20, 21 and 23). All of wells are rich in HCO3 and Ca and these wells emerged from limestone rocks, have abundant around of wells (see Fig. 2). Wells have Ca>Mg>Na and HCO3>Cl>SO4. Samples 13, 9, 11, 10, 3, 21 and 8 that have high Na and Cl belong to Ca-Na-HCO3-Cl type water show mixing of saline water with fresh water towards very low brackish (Fig. 6).

Figure 5 · Piper classification diagram of water samples
Figure 5
Figure 6 · Na versus Cl content of the studied wells
Figure 6
Figure 7 · Discriminant analysis for verifying factors
in factoring samples
Figure 7

4.3 Determining Major Hydrochemical Features using FA

The correlation matrix is shown in Table 3. The parameters that are highly related (r> 0.6) to each other including: EC with Ca, Cl, NO3 and HCO3; Ca with Cl and TDS; Mg with TDS; Na with Cl and SO4 (Table 3). Table 4 listed the eigenvalues of the first three factors, their percentage of variance and cumulative percentage of variance. It revealed that the eigenvalues of the three factors, which exceed one, explain 76.24 % of the total variance. Table 4 reported the loading of varimax normalized-rotated factor matrix for the three-factor model. The absolute values of factor loadings of over 0.7 were considered as strong correlation and marked in Table 4 to elucidate the relationships between the factors and the hydrochemical data.

To understand the principal role of the variables discriminating the three groups obtained by factor analysis, a discriminant analysis was applied to hydrochemical data of the study area. The qualitative–dependent variable consists of the classified groups of samples that resulted from factor analysis. Our aim was to verify if these groups were divided correctly by factor analysis. The discriminant analysis was 100 % successful, as all samples classified to the correct factor, thus indicating the feasibility of factor analysis (Fig. 7).

Table 3 · The Correlation Matrix of the Studied Wells (p <0.05000)
  pH EC Ca Mg Na Fe Cl SO4 NO3 F HCO3 TDS
pH 1                      
EC -0.43 1                    
Ca -0.29 0.68 1                  
Mg -0.19 0.46 0.22 1                
Na 0.01 0.56 0.44 0.21 1              
Fe -0.34 0.53 0.13 0.36 0.21 1            
Cl -0.12 0.74 0.74 0.40 0.65 0.27 1          
SO4 -0.12 0.51 0.54 0.16 0.68 0.46 0.66 1        
NO3 -0.45 0.85 0.58 0.44 0.56 0.59 0.69 0.64 1      
F 0.21 0.32 0.24 0.19 0.47 0.30 0.48 0.51 0.23 1    
HCO3 -0.49 0.60 0.45 0.51 0.03 0.56 0.42 0.38 0.53 0.30 1  
TDS -0.41 0.80 0.73 0.61 0.39 0.590 0.77 0.56 0.773 0.41 0.799 1
Table 4 · Factor Analysis of Groundwater Data from Arak City
  Factor 1 Factor 2 Factor 3
pH -0.43 -0.18 0.72
EC 0.53 0.72 -0.14
Ca 0.15 0.83 -0.22
Mg 0.68 0.14 0.04
Na -0.01 0.80 0.35
Fe 0.80 0.13 0.11
Cl 0.27 0.86 0.13
SO4 0.27 0.73 0.30
NO3 0.53 0.70 -0.15
F 0.32 0.34 0.74
HCO3 0.84 0.23 -0.16
TDS 0.72 0.61 -0.07
Eigenvalue 6.26 1.78 1.09
Total variance % 52.24 14.88 9.11
Cumulative-Eigenvalue 6.26 8.05 9.14
Cumulative % 52.24 67.13 76.24

Factor 1, which explains 52.24 % of the total variance, has strong positive loading on HCO3, Fe, TDS and has moderately positive loading on Mg. To compare with data on hydrochemical parameters of wells (Table 1), the HCO3, Fe and TDS item are the dominant solutes in well waters emerging from limestone formation. Factor 2 explains 14.88 % of the total variance with strong loading of Cl, Ca, Na, SO4 and NO3 (Table 4). To compare with hydrochemical data on wells (Table 1), the Cl, Na and SO4 items are the dominant solutes in wells water emerging from evaporite formation. Circulating of meteoric water contacting with saline rocks results in high Cl, Na and SO4 in the wells water (Chae et al., 2006). In other words, playa intrusion is a reason of increasing Cl, Na and SO4 concentration in wells water. Factor 3 explains 9.11 % of the total variance with strong loading of pH and F (Table 4).

To compare hydrochemical data together, it determined major anion and cation in soils over limestone in source rocks (4 samples ) and 5 samples from groundwater from wells close to Mighan playa in northeast of study area (see Fig. 2). Then, the results transferred in Piper diagram (Fig. 8). Groundwaters from wells close to saline playa are mostly of Na-Cl-SO4 type and soils in source rocks showing Ca-HCO3 while groundwater of study area is of mixed cation–anion. Therefore, in the first factor wells are close to limestone source rock rich in HCO3 and are fresh water. In the second factor, it is clear, the Cl, Na and SO4 items are significant. Wells are rich of above items have composition similar to wells that are close to saline Mighan playa (Fig. 9) and the quality of waters is very brackish for drinking. In second factor, nitrate is also significant and it has anthropogenic origin. Fluorine may essentially be from a natural origin (third factor). Limestone formations can contain significant quantities of fluorine which can be liberated by the water-rock interaction (Ketata et al., 2011).

Figure 8 · Piper diagram showing the compositions of groundwater in study area, groundwater close to Mighan playa and soils in source rocks
Figure 8
Table 5 · Factor scores, based on groundwater hydrochemical data
Factor 1 Factor 2 Factor 3
1 -0.63 0.13 0.19
2 -1.32 -0.40 0.03
3 0.10 1.44 -0.23
4 3.15 -1.96 0.65
5 0.21 -0.09 -0.69
6 0.32 -0.73 0.55
7 1.66 -0.36 -1.27
8 -0.77 0.36 1.10
9 0.07 1.37 -1.01
10 0.42 0.66 -0.51
11 0.94 1.95 1.07
12 0.66 0.82 -1.02
13 -0.10 2.39 -1.17
14 0.62 0.036 -0.27
15 0.36 -0.60 -0.13
16 0.53 -0.74 0.05
17 0.25 0.38 1.
18 0.80 0.02 0.13
19 0.05 0.80 0.01
20 0.25 -0.23 2.17
21 -1.15 0.25 1
22 -0.56 -0.54 0.981
23 -1.16 -0.05 0.92
24 -0.78 -0.42 1.03
25 -1.35 -1.12 0.14
26 -1.21 -1.08 -0.98
27 -1.07 -0.99 -2.08
28 -0.32 -1.29 -1.69

4.4 Factor Scores

The factor score were also estimated to find out the spatial variation of the factor representation and to identify the zone of its representation (Reghunath et al., 2002; Lambrakis et al., 2004). They are commonly obtained by weighted least square method (Table 5). The positive zones indicate the dominance of that factor. The spatial variation by using the factor score values of each sampling points were plotted by Surfer Software Version 10. The spatial representation for first factor with higher cretaceous limestone region where the influence of lithology is preferably noted (Fig. 9a). This suggests that HCO3 water ingress into the aquifer was predominantly related to water recharge from limestone formation in west of studied area. The high score area is the most densely freshwater region. The spatial representation of second factor (Fig.9b) is mainly represented in the north–east of studied area. They represent leaching of salts and anthropogenic impact. The higher scores of third factor is represented in the regions of agricultural practices (Fig .9c). In the north east side, saline water intrusion into the aquifers, and anthropogenic activities from waste disposal practices are the major controlling factors in the study area.

Figure 9 · Contours of factor scores, showing the geographical distribution of the first factor (9a), second factor (9b) and third factor (9c) in the studied area
Figure 9

5. Conclusion

The salinity of groundwater from the aquifers of Arak city has been determined to be controlled by two major factors: dissolution mineral in limestone formation, and saline playa intrusion and anthropogenic effects. Two distinct categories of groundwater have also been identified: moderate EC, moderate pH waters which are suitable for most uses and identified mainly close the source rocks; intermediate high EC groundwaters which are moderate suitable for most domestic purposes on account of moderate concentrations of nitrate identified. The quality of groundwater has been observed to decrease towards the Mighan playa, largely due to playawater intrusion. This study finds that the major water type in the aquifers of Arak is the Ca-HCO3 water type, which gradually degrades into very low saline Ca-Na-HCO3-Cl water types towards the Mighan playa.

From the elaboration of a set of hydrochemical data, factor analysis identified three factors that explained 76.24 % of the total variance. The first factor shows the relationship between HCO3, Fe, TDS concentrations and dissolution of carbonate rocks such as limestone formation or natural origin. The second factor shows relationship between Na, CL, SO4, Ca and NO3 concentration and intrusion playawater, anthropogenic factors and in the third factor, F, pH shows natural factor. Geographical distribution of scores of factors identifies areas with higher second factor, concentrated in northeast of area and towards Mighan playa.

Acknowledgements

The authors would like to thank the Department of Mining Engineering of Arak University of Technology for the hydrochemical analysis.

References

***

Copyright © 2012, ECO Services International