Using correlation and multivariate statistical analysis to identify hydrogeochemical processes affecting the major ion chemistry of waters: a case study in Laoheba phosphorite mine in Sichuan, China
- First Online:
- Cite this article as:
- Wu, J., Li, P., Qian, H. et al. Arab J Geosci (2014) 7: 3973. doi:10.1007/s12517-013-1057-4
Prior to mining, the water in and around the mine is rarely influenced by human activities, and hydrogeochemical processes are the major factors influencing and controlling water chemistry. To identify these natural hydrogeochemical processes in Laoheba phosphorite mine (Sichuan Province, China), correlation and multivariate statistical techniques were used. Results show that water quality in the area is generally good before the Laoheba phosphorite mine goes into construction and production. The cluster analysis classified water samples into 4 clusters (C1–C4). Samples from C1 and C2 are of HCO3−Ca·Mg and HCO3−Ca type, while those from C3 and C4 are of HCO3−Ca·Mg type. Most parameters except Cl− and pH show an increasing trend in the order of C1 to C4. Three principal components were extracted, and PC1 represents the ion exchange and the weathering of calcite, dolomite, and silicate minerals. PC2 and PC3 indicate the process of water recharge from upstream waters and the process of evaporation, respectively. The hydrochemistry of waters in the area is a result of multiple factors, and natural mineral weathering and ion exchange are the most important ones.
KeywordsHydrogeochemistryGroundwater resourcesMajor ion chemistryWater–rock interactionMineral weathering
In mining areas, water resources are important for human consumption, plant growth, and mine exploitation and their rational development has always been an important issue worthy of discussion. In mining areas, heavy metal pollution in water, soil, and sediments is popular and many cases of this kind have been reported (Prasad and Bose 2001; Razo et al. 2004; Nikolaidis et al. 2010). In fact, in addition to heavy metal pollution, various mining activities can also influence the concentrations of major ions in surface and groundwater in and around a mine (Singh et al. 2011). Therefore, it is quite important and useful to investigate and understand the major ion concentration variations prior to and after the mine production.
Generally, prior to mining, major ions in water are usually controlled by natural processes, such as mineral weathering, ion exchange, water–rock interactions, etc. (Li et al. 2013a). These natural processes are slow (Li et al. 2013a; Qian et al. 2012). However, after the mine goes into construction and production, human interference may increase the speed of these processes, resulting in the noticeable variation of these major ions. Technically, water resources protection in mining areas requires the understanding of water quality status and processes that control the major ion chemistry before it is seriously polluted. As a result, major ion chemistry investigation before water pollution is an important and necessary task for water resources protection in mining areas.
Research on water quality and hydrochemistry has been widely conducted over the world due to the increase of awareness of water quality protection. Arnous and El-Rayes (2012) developed a GIS-based model to assess groundwater contamination in the West Ismailia area. They produced a groundwater potential risk map and reported that the groundwater pollution may result from agricultural, domestic, and industrial activities. Similarly, Alfy (2012) assessed groundwater contamination integrating geostatistics and GIS techniques in Al Arish area, Sinai, Egypt. Dar et al. (2012) found nitrate and total hardness pollution in some villages in Tamil Nadu after an investigation on environmental chemistry of groundwater in this area. Hydrochemical investigations have also widely been carried out in various mining areas. Rozkowski et al. (1993) have realized the negative effects of mining on groundwater chemistry and environment in the early 1990s. Mine water usually contains high concentrations of chemical constituents such as sulfate, fluoride, and heavy metals released from surrounding rocks because of natural processes and human interferences (Makni et al. 2013; Aleksander-Kwaterczak and Ciszewski 2012). Janson et al. (2009) studied the quality of mine water in a coal mine, and Aleksander-Kwaterczak and Ciszewski (2012) investigated groundwater and soil pollution near an abandoned lead–zinc mine. Favas et al. (2012) and Byrne et al. (2013) reported the effects of mining activities on the quality and hydrochemistry of surface water. These studies have shown that mining activities will have significant influences on surface water and groundwater as well as soils surrounding the mine.
The Laoheba phosphorite mine, located in Mabian County, Sichuan Province, belongs to Zhongyi Mining Development Co. Ltd. After the mine goes into production, mining activities will probably change the hydrochemical characteristics and quality of waters in and around the mining area. Nevertheless, before anthropogenic disturbance, natural hydrogeochemical processes such as mineral weathering and ion exchange probably are the most important factors affecting the water chemistry. Therefore, the present study aimed at investigating the main hydrogeochemical processes controlling and influencing the water chemistry in and around the Laoheba phosphorite mine using correlation and multivariate statistical techniques.
Materials and methods
The study area belongs to the tropical monsoon climate zone. Warm and moist air from the Sichuan Basin is blocked by the high mountain at the western edge, producing a foggy and rainy climate in the area. According to the data from Mabian weather station, the local annual average temperature is 16.9 °C, the highest average temperature is 25.2 °C, and the lowest average is –2.3 °C (Xiang and Zhang 2011). Rainfall is abundant and the annual rainfall is 1,052.3 mm. August has the highest precipitation (256.4 mm) and January the lowest (4.2 mm). The area is part of the Yangtze River drainage area, and the surface water resources are well developed. There are mainly three rivers running from west to east (Fig. 1). The three rivers are all tributaries of the Mabian River. The central river near Shanshuigou has the highest flow rate with a maximum value of 3,200 L/s and an average of 950 L/s. The river near Kucaiwan is the smallest with the maximum and average flow rates of 50 and 17.5 L/s, respectively. In many places of the area, groundwater is usually discharged as springs, accounting for a great proportion of river flow. Therefore, close physiochemical relationship between surface water and groundwater can be expected.
Aquifer media in the area is mainly composed of carbonate minerals formed in Cambrian, Sinian, and Quaternary periods. Therefore, karst water, pore water, and fissure water are popular here. Groundwater generally flows from northwest to southeast and the water chemistry types are HCO3−Ca·Mg and HCO3−Ca. Groundwater gets recharge from precipitation through fissures.
Sample collection and analysis
Physicochemical index of water samples
Where all cations and anions are expressed in milliequivalent per liter. The results show that most samples have a %CBE smaller than ±5 % except sample G11 which possesses a %CBE of 10.038 %. Therefore, sample G11 was not used in the study.
Multivariate statistical techniques
Multivariate statistical techniques such as principal component analysis (PCA), cluster analysis (CA), and discriminant analysis are popular methods for solving multivariate problems (Tang 2010). They have been widely used in geology, hydrology, meteorology, medical science, industry, agriculture, and environmental science (for example, Baeza and Corominas 2001; Cloutier et al. 2008; Koklu et al. 2010; Yidana et al. 2010). In the present study, PCA and CA were selected and applied in hydrogeochemical data analysis. CA is a data classification technique which groups samples or indices with similar characteristics, while PCA is a data transformation technique that reveals a simple underlying structure within a multivariate dataset (Cloutier et al. 2008).
Where cluster r is clustered by sub-clusters p and q, d represents the distance between different clusters, αp, αq, β, and γ are coefficients. In furthest neighbor method, αp = αq = γ = 1/2, β = 0.
Where y1, y2, …, yn are independent variables subject to the condition that ck12 + ck22 + … + ckp2 = 1(k = 1, 2, …, p).
Results and discussion
TDS and TH are in the range of 96.68 to 242.52 mg/L and 92.07 to 223.74 mg/L, respectively (Table 1), indicating fresh and soft water. These values generally show a gradual increasing trend toward southeast, which implies that more minerals have dissolved during water flow. Accordingly, the saturation indices of major minerals also show a rough increasing trend. Saturation index of calcite is in the range of −0.54 to 0.26, suggesting a general equilibrium state, while those of dolomite and gypsum range from −3.44 to −0.03 and from −2.65 to −1.77, respectively, implying unsaturated state of the two minerals. With water flowing, more dolomite and gypsum can dissolve in the water (Li et al. 2013b). The pH values of groundwater and surface water are in the range of 7.40 to 8.03, indicating alkalescent water.
Correlation of parameters
Pearson correlation of physiochemical parameters
However, the ratio between Ca2+ and SO42− is not theoretical 1:1. Taking into account the dissolution of dolomite which will introduce Ca2+ into water, it is logical to conclude that the deviation of plots away from the theoretical 1:1 line is caused by calcium-containing mineral weathering such as calcite and dolomite. The three different water bodies (surface, spring, and mine water) have similar variation trend and/or relations between ions, which indicates similar processes have occurred in these waters. However, in addition to natural processes of mineral weathering and water–rock interactions, water chemistry may also be influenced by other factors such as recharge water chemistry, precipitation, or evaporation.
Q mode CA is usually used to highlight spatial relationships among the sample points, while R mode CA is done to classify the parameters into groups or facies based on their similarity with each other (Banoeng-Yakubo et al. 2009). In the study, Q mode CA was performed with SPSS 13.0 for Windows. A combination of the Euclidean as a similarity measure and the furthest neighbor agglomeration scheme to link clusters has been determined to yield optimal results in the CA. The samples with the larger similarity were first grouped, and then groups of samples were joined with a linkage rule. The steps were repeated until all observations had been classified.
Median concentrations of parameters of different clusters
The stiff diagrams (Fig. 4) and the data (Table 3) indicate that the four clusters are geochemically independent groups. Samples from C1 and C2 are of HCO3−Ca·Mg or HCO3−Ca type and the concentrations of all ions except Cl− are lower than those of samples from C3 and C4. Besides, TH and TDS are also lower in C1 and C2 than in C3 and C4. Samples from C3 and C4 are of HCO3−Ca·Mg type. They have more HCO3− and Ca2+ indicated by the Stiff diagrams. Most parameters except Cl− and pH show an increasing trend in the order of C1 to C4. The Cl− concentration and pH do not have a noticeable variation trend, which is probably caused by its small variation in values. Samples of C1 have the lowest ion concentrations, which may indicate that they have not undergone long residence within the geosphere, and their interactions with rocks are not sufficient. This is confirmed by the saturation index listed in Table 1. Samples from C1 usually have small saturation index values which suggest that they are unsaturated. On the contrary, samples from C4 have higher values of most ions, which may be explained by longer residence time and sufficient interactions with rocks. The saturation indices of minerals in C4 are usually higher than those of samples in C1.
Total variance explained by each PC and the loading matrix of PCs
% of Variance
PC1 explains the greatest of the variance (51.10 %) and is characterized by highly positive loadings in TDS, TH, HCO3−, Ca2+, Mg2+, CO2, and Na++K+ (Table 4), which represents the ion exchange and the weathering of calcite, dolomite, and silicate minerals as suggested by correlation analysis. It is a major factor influencing the water chemistry. PC2 and PC3 explain 18.48 and 11.87 % of the variance, respectively, indicating that these components are related to more local effects than PC1, and they are secondary factors. PC2 is highly correlated with CO32− and pH, which may represent the process of water recharge from upstream waters which usually have relative higher pH values, while PC3 is positively correlated with Cl− and negatively correlated with SO42−, which may be explained by the process of evaporation since evaporation is great in this area.
Before the Laoheba phosphorite mine goes into construction and production, the water quality is good and fit for drinking with regard to the major physiochemical parameters. It is controlled by natural weathering processes and influenced by recharge water quality and evaporation indicated by correlation analysis.
Four geochemically distinct clusters can be classified by CA (C1–C4). Samples from C1 and C2 are of HCO3−Ca·Mg and HCO3−Ca type, and those from C3 and C4 are of HCO3−Ca·Mg type. Most parameters except Cl− and pH show an increasing trend in the order of C1 to C4.
The first three PCs explain 81.46 % of the total variance in the dataset. PC1, characterized by great positive loadings in TDS, TH, HCO3−, Ca2+, Mg2+, CO2, and Na++K+, represents the ion exchange and the weathering of calcite, dolomite, and silicate minerals. PC2 and PC3 indicate the process of water recharge from upstream waters and the process of evaporation, respectively. The hydrochemistry of waters in the area is influenced by multiple factors, and mineral weathering and ion exchange are the most important ones, but the influences of recharge water quality and evaporation cannot be ignored.
The research was supported by the Doctor Postgraduate Technical Project of Chang'an University (2013G5290002 and CHD2011ZY022), the Special Fund for Basic Scientific Research of Central Colleges (CHD2011ZY020 and CHD2012TD003), and the National Natural Science Foundation of China (41172212). We are grateful to the editor and reviewers for their valuable comments. Engineer Xiang G was highly appreciated for providing the basic physiochemical and hydrogeological data of the study.