Influences of natural and anthropogenic processes on the groundwater quality in the Dagujia River Basin in Yantai, China

Groundwater is a vital provision resource for agriculture, industry, public consumption and rural communities (Ansa-Asare et al. 2009). The quality and quantity of groundwater are greatly dependent on environmental and anthropogenic factors, such as regional geological setting, precipitation and human activities (Devic et al. 2013; Hamzah et al. 2017). Some anthropogenic activities, such as urban development, agricultural production, industrial application, mining, power generation and forestry practices, have caused the deterioration in groundwater quality and quantity (Jiang et al. 2009; Khatri & Tyagi 2015; Khatri et al. 2017; Habib-ur-Rehman et al. 2018). Especially, high abstraction rate or overpumping for the coastal aquifers might lead to seawater intrusion, which could increase the levels of Na, Cl and total dissolved solids (TDS) in the groundwater (Rao et al. 2013). The high ionic concentration could limit the groundwater usage. Therefore, it is very important to understand the influencing/controlling factors of groundwater quality for the groundwater management and utilization (Nishikawa 2016).

As known, the groundwater quality largely depends on the natural processes (such as lithology, hydrogeological conditions, the interaction of water with rock, soil and other types of aquifers, and seawater intrusion), anthropogenic factors (such as agriculture, industry and urban development) and atmospheric input (Jeong 2001; Jiang et al. 2009; Huang et al. 2013). Many previous studies have reported the influences of natural and anthropogenic processes on the groundwater quality in the coastal areas (Kim et al. 2003; Xing et al. 2013), arid/semi-arid areas (Jianhua et al. 2012; Wu et al. 2017; Xiao et al. 2018; Yang et al. 2018) and karst area (Jiang et al. 2009). The relative results suggested that anthropogenic processes played an important role in determining the groundwater chemical components. Generally, the application of water indices analysis, such as the hierarchy method based on the index system (Zhu et al. 2018) and the criticality index method (Pimparkar et al. 2016), and multivariate statistical techniques, such as hierarchical cluster analysis (HCA) and factor analysis (FA) (Alberto et al. 2001; Reghunath et al. 2002; Simeonov et al. 2004; Omo-Irabor et al. 2008; Hynds et al. 2014), allows us to better understand the groundwater quality, identify the possible factors and finally offer a valuable tool for managing water resources and rapid solution to pollution problems. For example, the FA method has been widely used to assess and characterize surface water and groundwater, which is useful in identifying the variations induced by natural and anthropogenic factors (Liu et al. 2003; Shrestha & Kazama 2007; Jiang et al. 2009).

As a typical coastal area, the Dagujia River Basin is located in the middle-north of Yantai city along the Dagujia River, which is one of the most important freshwater resources. Several studies on the evolution of groundwater quality with different technical methods have focused on the Dagujia River Basin (Guo 2005; Li & Liu 2007). However, the previous studies mainly focused on the basic characteristics of groundwater quality, lacking of the analysis of factors affecting the groundwater quality and the relevant affecting mechanisms.

On the basis of analyzing concentrations of the conventional pollutants and minor metal elements, the objectives of this study are to (1) describe the physico-chemical characteristics of the groundwater in the study area and (2) distinguish the impact of natural factors and human activities on the groundwater quality with FA and HCA methods. This study would provide important data and a theoretical basis for groundwater management and utilization and provide references for other coastal areas in China.

The study area, named the Dagujia River Basin, is located in the middle-north of Yantai city in China and covers an area of about 2,435 km² (Figure 1). It lies between latitudes 37°2′31.478″–37°37′51.592″ N and longitudes 120°44′1.857″–121°26′27.636″ E. The Dagujia River is the second largest river of Yantai and occupies an important position in the water supply and economic development of Yantai city. Geomorphologically, the area is composed of about 40% low mountains, 40% hills and 20% valley plain. The annual mean temperature of the area is about 11.5 °C (Liu et al. 2007). The main way of groundwater recharge is atmospheric precipitation, and the annual average precipitation is 677.95 mm (Zeng et al. 2017).

The Quaternary strata are the main characteristics of the Dagujia River Basin, and the groundwater source is mainly alluvial sand and sand–gravel pore water with the total area of 63.3 km² (Tao et al. 2018). Up to now, five groundwater resources have been exploited. The hydrogeological map in the previous study showed that from bottom to top, the three-layer strata are in the following order: confined aquifer (mainly, sand, gravel and pebble), relative water-blocking layer (silty sand and silt) and porous aquifer (fine silt, silt and a part of medium sand) (Tao et al. 2018). Due to the difference of topography, stratigraphic lithology, runoff condition and distance from the sea, the hydrochemical type of the groundwater changes from HCO₃–Ca·Mg to Cl–Na from south to north in the Dagujia River Basin (Kou 2013).

Forty-six groundwater samples, including 7 confined samples and 39 phreatic samples, were collected from different wells in 22–27 May 2014, and the location of the wells is illustrated in Figure 1. The sampling depth ranges from 1.7 to 100 m. Before the samples were collected, the wells were pumped out for about 30 min to remove stagnant or polluted water. The collected samples were stored at 4 °C. The selected groundwater parameters include pH, electrical conductivity (EC), turbidity, alkalinity (Alk), TDS, total hardness (TH), Ca²⁺, K⁺, Na⁺, Mg²⁺, Cl⁻, SO₄²⁻, HCO₃⁻, F⁻, NO₃⁻, As, Pb, Cr, Hg, Zn, Cd, Se, Mn, Fe and Al. The indices pH, EC and turbidity were measured in the field, and other indices were analyzed in the laboratory. The physico-chemical analyses of all samples were performed according to the National Sanitary Method for Determination of Drinking Water (GBT 5750-2006), and the related analysis methods are listed in Table S1 in Supplementary Materials.

As shown in Table S2, the groundwater type of wells 16, 17, 38, 39, 40, 42 and 44 is confined water, while that of other wells is phreatic water. The three wells 2, 17 and 44, five wells 16, 29, 33, 38 and 41 and the other wells are screened in the carbonate, igneous and metamorphic, and loose rock aquifers, respectively. It also can be seen that from Table S2, the utilization of well water is mainly for drinking, agricultural irrigation, industry and domestic water.

Finally, the data were transformed into factors, and the eigenvalue and the percent of variance associated with each factor were obtained. The number of factors extracted is determined in a way that reduces the data by using the inherent interdependences. Therefore, a small number of factors could be remained and account for about the same amount of all original data.

HCA, one of the most commonly used multivariate statistical methods, could group the factors into clusters on the basis of similarities within a cluster and dissimilarities between different clusters (Zhang et al. 2012; Huang et al. 2013; Abu-Alnaeem et al. 2018). The results of HCA can help in interpreting the hydrochemical data and patterns.

The selected variables for FA and HCA were pH, EC, TDS, Ca²⁺, K⁺, Na⁺, Mg²⁺, Cl⁻, SO₄²⁻, HCO₃⁻ and NO₃⁻.

The spatial distribution of factors (Figure 2) is derived from space interpolation analysis with the inverse distance weighting method.

Fifteen physico-chemical parameters were detected and are shown in Tables 1 and 2. The pH of the groundwater samples varies from 6.84 to 8.07. The majority of the groundwater samples is moderately mineralized, suggested by its moderate to high EC values from 549 to 4,000 μS/cm. The highest EC value is found at both wells 5 and 9, which is probably ascribed to their short distances to the coast. A wide range in TDS concentrations from 396.92 to 19,436.22 mg/L with an average of 1,141.45 mg/L is recorded for all the groundwater samples. Notably, the maximum concentrations of TDS (19,436.22 mg/L), TH (3,676.21 mg/L), K⁺ (172.5 mg/L), Na⁺ (5,875 mg/L), Ca²⁺ (306.69 mg/L), Mg²⁺ (706.75 mg/L), Cl⁻ (1,0684.75) and SO₄²⁻ (1,575.19 mg/L) are all found in the well 5, which is closest to the coastal line in the north, suggesting that these values are probably related to seawater intrusion.

Concentrations of minor elements (As, Pb, Cr, Hg, Zn, Cd, Se, Mn, Fe and Al) were also detected and are displayed in Table S3. As expected, the concentrations of minor elements in most groundwater wells are low. The relatively high pH (>6.5) in the majority of the groundwater samples reduces the solubility of most minor elements, whose solubility in water strongly relies on pH (Appelo & Postma 2005; Lin et al. 2011). However, Mn and Al in several samples showed the elevated concentrations. The concentrations of Mn in wells 8, 14 and 20 are 0.58, 1.26 and 0.61 mg/L, respectively, which all exceed the WHO limits (0.4 mg/L) in Guidelines for Drinking-Water Quality (WHO 2009), and Mn values in four well samples (wells 6, 8, 14 and 20) are higher than the limit value (0.3 mg/L) in the Drinking Water Sanitary Standard (GB 5749-2006). Additionally, the elevated concentration of Al (0.26 mg/L) is found in only one well sample (well 2), which indicates that the well 2 is not suitable for drinking.

Variables for FA in the study were pH, EC, TDS, K⁺, Na⁺, Ca²⁺, Mg²⁺, Cl⁻, SO₄²⁻, HCO₃⁻ and NO₃⁻. The correlation matrix for the 11 parameters is displayed in Table 3. As seen from Table 3, high correlation coefficients are found between most of the corresponding two variables, indicating that the selected variables could be classified by FA. The eigenvalues, the percent of variance, the cumulative eigenvalue and the cumulative percentage of variance associated with each other are shown in Table 4. Generally, the maximum number of factors is extracted based on the Kaiser criterion (Jiang et al. 2009), taking factors only into account with eigenvalues higher than 1. Accordingly, three factors were selected and then rotated based on the Varimax with the Kaiser Normalization method. The corresponding factor loadings and communality of the variables for the three factors are listed in Table 5.

From Table 4, it can be noted that the first three factors account for about 85.09% of the total variance, which indicates that it is reasonable and suitable to extract the first three factors as the main factors. As listed in Table 5, the communalities of all variables except pH and HCO₃⁻ were higher than 0.80. The different factor loadings in Table 5 suggest different contributions in identifying the chemical compositions of the groundwater samples.

The terms ‘strong’, ‘moderate’ and ‘weak’ were used to denote the absolute factor loading values of >0.75, 0.75–0.50 and 0.50–0.30, respectively (Liu et al. 2003). Factor 1 consists of EC, TDS, K⁺, Na⁺, Ca²⁺, Mg²⁺, Cl⁻ and SO₄²⁻, accounting for about 63.12% of the total variance. As known, EC, TDS, K⁺, Na⁺, Ca²⁺, Mg²⁺, Cl⁻ and SO₄²⁻ are the primary parameters in seawater (Liu et al. 2003). The previous study by Zeng et al. (2017) showed that the increase in the Cl⁻ concentration was in positive correlation with the seawater intrusion in the Dagujia River Basin. Moreover, as seen from Table 3, EC is positively connected with TDS, K⁺, Na⁺, Ca²⁺, Mg²⁺, Cl⁻ and SO₄²⁻. Therefore, EC can be noted as a water salinization index. The association of EC, TDS, K⁺, Na⁺, Ca²⁺, Mg²⁺, Cl⁻ and SO₄²⁻ could reflect the influence of seawater intrusion on groundwater pollution (Liu et al. 2003). Factor 1 can be termed as ‘the seawater intrusion factor’.

The previous studies showed that overpumping of groundwater was the main cause of seawater intrusion (Lambrakis et al. 1997; Liu et al. 2003). In this study, it can be seen from Figure 2(a) that the wells with high scores, shown as the dark colors, are likely distributed in the north of the study area, which is closer to the coast. Coincidently, the areas with dark colors are the main residential and agricultural lands (shown in Figure S1). The finding confirms that overpumping of groundwater for domestic water and agricultural water is the major cause of seawater salinization in the study area. Especially, the highest score at well 5 suggests heavy seawater salinization. The spatial distribution of Factor 1 (Figure 2(a)) indicates that the well locations are corresponding to the seawater intrusion.

Factor 2, explaining 11.71% of the total variance, is composed of pH, EC and HCO₃⁻. The ‘strong’ positive loading of HCO₃⁻ indicates that the variable is related to the water–rock interaction. The negative loading of pH and positive loading of EC also support the water–rock interaction hypothesis (Jiang et al. 2009). The EC value reflects the amounts of dissolved ions, and the pH of groundwater denotes the H⁺ concentration in groundwater. However, H⁺ could be absorbed by the water–rock interaction process (Pacheco & Weijden 1996). Therefore, Factor 2 is assumed to be termed as ‘the water–rock interaction factor’.

Factor 3, accounting for about 10.23% of the total variance, mainly includes NO₃⁻. As known, no lithologic source was found for NO₃⁻, and atmospheric deposition was not generally regarded as a main source of NO₃⁻ in the groundwater (Jiang et al. 2009). Generally, NO₃⁻ could be introduced into the groundwater by urea fertilization in farmlands. The extra amount of fertilizer could not be absorbed and quickly degraded in the soil and then could migrate down to the groundwater (Jiang et al. 2009). Moreover, NO₃⁻ could be introduced by the sewage generated from the residential areas. Hence, the high concentration of NO₃⁻ in the study area was probably derived from the anthropogenic processes. Noticeably, the land-use type (Figure S1) around the wells with high NO₃⁻ concentrations is mainly agricultural land and urban land, and the surroundings possibly affecting the groundwater quality (Table S2) are some livestock farms and farmlands, which could provide an evidence for that anthropogenic processes could cause high NO₃⁻ concentration in groundwater. Therefore, the Factor 3 is assumed to be the indicative of human activities.

To emphasize the relation of the mentioned influences of seawater intrusion, geochemical processes and human activities on the chemical compositions of groundwater, the contribution of each factor at each well (factor scores) is analyzed and listed in Table 6, and the corresponding maps are illustrated in Figure 2.

As illustrated in Figure 2(a), the dark degree denotes for the score of Factor 1 for different wells, and the area with darker color represents a higher pollutant concentration included in Factor 1. The wells with high scores, shown as the dark colors, are likely distributed in the north of the study area, which is closer to the coast. Especially, the highest score at well 5 suggests heavy seawater salinization. It can be obtained that the distance to the sea probably affects the score of the wells, and thus the spatial distribution of Factor 1 indicates that the well locations are corresponding to the seawater intrusion. Figure 2(b) gives the geographical distribution of Factor 2 for all the 46 wells. The high scores at wells 8, 13 and 40, which are the darkly shaded areas, indicate the strong water–rock interaction. Figure 2(c) shows three wells 7, 25 and 39 with high scores, indicating that high concentrations of NO₃⁻ are detected in these groundwater samples. It can be seen from the land-use map (Figure S1) that the land types around these wells are mainly agricultural and residential lands, which are in good accordance with the spatial distribution of Factor 3. The distribution of the comprehensive factor (F) (Figure 2(d)) displays that the highest score with the darkest color is found at well 5, suggesting that the groundwater quality of the well 5 is most affected by the natural and anthropogenic processes. The distribution of different factors is in good accordance with the groundwater physico-chemical quality of different wells.

Except the FA method, the HCA method was also proved to be effective to cluster the variances based on the chemical similarity. The wells with similar hydrochemical characteristics were classified into the same group. The results using HCA are shown in Figure 3.

As illustrated in Figure 3, three clusters were obtained by selecting the phenon line with a linkage distance of 13. The selection of the phenon value is on the basis of yielding the lowest number of groups which could elucidate most satisfactory results of the difference in chemical characteristics (Cloutier et al. 2008; Monjerezi et al. 2011). Cluster 1 includes parameters of Na⁺, Cl⁻, TDS, Mg²⁺, SO₄²⁻, K⁺, EC and Ca²⁺, which are exactly as the terms contained in Factor 1. Therefore, Cluster 1 could be the indicator of the seawater intrusion. The parameters pH and HCO₃⁻ are classified as Cluster 2. The difference between Cluster 2 and Factor 2 is that EC is not included in Cluster 2. As mentioned, EC has a ‘weak’ positive loading for Factor 2, which indicates that parameters pH and HCO₃⁻ are the main factors affecting the Factor 2. Since factors HCO₃⁻ and pH are concerned in the water–rock interaction (Lin et al. 2011; Huang et al. 2013), the Cluster 2 could be termed as ‘the water–rock interaction factor’. As Factor 3, Cluster 3 is comprised of NO₃⁻, which is mainly generated from anthropogenic processes (fertilizer application and domestic sewage discharge) (Huang et al. 2013). Accordingly, Cluster 3 could be assumed to be the indicative of human activities. It can be noted that the results from HCA are generally consistent with those from FA, suggesting that both of the two methods are valuable tools to understand the possible pollution sources of groundwater.

The physico-chemical characteristics of groundwater in the Dagujia River Basin were studied to clarify the situation of groundwater quality and speculate the possible factors affecting the groundwater quality. A mass of 46 groundwater samples were collected and analyzed, and the results showed that several groundwater quality indices in some wells (such as wells 2, 6, 8, 14 and 20) are out of WHO limits and the standards of GB 5749-2006, indicating that these wells are not suitable for drinking. A great variation of chemical components was found across the study area, where the wells with high TDS, TH, K⁺, Na⁺, Ca²⁺, Mg²⁺, Cl⁻ and SO₄²⁻ concentrations were inclined to be located close to the coastal zone. The FA and HCA results showed that seawater intrusion was the primary mechanism controlling the groundwater quality in the coastal areas. A three-factor model was proposed based on FA and explained over 85% of the total groundwater quality variation. Factor 1 was denoted as ‘the seawater intrusion factor’, consisting of EC, TDS, K⁺, Na⁺, Ca²⁺, Mg²⁺, Cl⁻ and SO₄²⁻ and accounting for 63.16% of the total variance. Factor 2 mainly included pH, EC and HCO₃⁻, explaining about 11.71% of the total variance, which was thought to be related to the water–rock interaction. Factor 3 (NO₃⁻) was assumed to be the indicative of human activities. Additionally, the geographical maps of the factor scores clearly illustrated the spatial distributions of wells affected by natural processes or anthropogenic processes. Three clusters were also generated from HCA, which were generally consistent with the results from FA. The results showed that both natural processes and human activities are the major factors affecting the groundwater quality. The results could provide references for the local government to protect groundwater ecology and allocate water resources.

This work is supported by the Shandong Province Geological exploration projects (No. 201345).

The Supplementary Material for this paper is available online at http://dx.doi.org/10.2166/aqua.2019.113.