Hydrochemistry of groundwater from loose layer aquifer system in northern Anhui Province, China: source of major ions and hydrological implications

It has become evident in many countries of the world that groundwater is one of the most important natural resources (Zektser & Lorne 2004). As a source of water supply, groundwater has a number of essential advantages when compared with surface water, such as higher quality, less subject to seasonal fluctuations, and much more uniform spread over large regions than surface water. These advantages have resulted in wide groundwater use for water supply. For example, groundwater is the only source of water supply for some countries in the world (Denmark, Malta, Saudi Arabia, etc.). Moreover, in countries with arid and semi-arid climates, about one-third of the landmass is irrigated by groundwater. Groundwater is also one of the most important natural resources in the North China Plain (Chen et al. 2005). It provides ∼56% of the water supply for more than 100 million people. Groundwater is also the source of much of the water used for irrigation (Zhang et al. 2000).

However, for the coal mines, groundwater is a double-edged sword: it is important for the human activities in the coal mining areas. The groundwater can be used for drinking, irrigation, and industrial purposes, especially under the condition of surface water shortage (Sun & Gui 2013). However, it is also a threat to the safety of coal mining, because water is considered to be the most dangerous of the five typical disasters in coal mines (water, fire, gas, dust, and roof), because water inrush has produced the highest loss to humans, not only the loss of property, but most importantly, the death of people (Gui & Chen 2007).

As one of the most important tools, hydrochemistry has played a significant role for controlling water disasters in coal mines, because it can be used for water source identification, which is essential for water disaster prevention and management. Consequently, a large number of studies related to groundwater hydrochemistry have been produced, and most of them have focused on statistics (e.g., Jiang & Liang 2006; Chen et al. 2009; Zhang et al. 2009; Zhou et al. 2010; Sun & Gui 2012). However, the mechanism regarding water–rock interaction in the groundwater system (such as the source of chemical constitutes) has not been well understood, which limits the popularization and application of these methods.

Groundwater from the loose layer aquifer system is considered to be the most important water resource in the North China Plain (Foster et al. 2004). It is also important for the groundwater supply in northern Anhui Province, China, which is an important area for the production of coal in China, with more than 100 million tons of coals being produced per year. It is also a threat to the safety of coal mining activities, because it can recharge the coal bearing aquifer (the aquifer faced by coal mining) through faults or other channels (Sun & Gui 2015).

In this study, a total of 43 groundwater samples from the loose layer aquifer system in Huaibei coalfield, northern Anhui Province, China have been collected, and their major ion concentrations measured and analyzed by EPA Unmix model. The goals of the study include: (1) understanding the chemical compositions of the groundwater from the aquifer system; (2) identifying and quantifying the sources of the major ions in the groundwater; and (3) a preliminary understanding of the mineral compositions of the aquifer system, which cannot be simply obtained by drilling programs because of the spatial inhomogeneity.

The Huaibei coalfield is located in the eastern part of Yu-huai depression zone, the southeast margin of North China Craton. The coalfield is bounded by the Guzhen–Changdeng fault in the east, the Guangwu–Guzhen fault in the south, the Xiafu–Gushi fault in the west, and the Fengpei fault in the north. The recharge, flow, and discharge of the groundwater are controlled by the faults around the coalfield, and the field is a closed to semi-closed grid hydrogeological unit.

The loose layer strata in the field cover the Permian coal bearing strata (Figure 1), and the thickness of the loose layer strata ranges from 80.5 to 867 m and most of the area is about 350 m. There are four aquifers and three aquicludes developed in the loose layer aquifer system. Owing to the existence of the third aquiclude, the impacts for water inrush during coal mining from the first, second, and third aquifers from shallow to deep are limited.

However, the distribution of the fourth aquifer is wide with a thickness between 0 and 59.1 m, and the wall rocks in the aquifer include conglomerate, sandstones, and clays. The amount of water in the aquifer is medium. During coal mining, the groundwater in the aquifer can flow into the coal bearing strata through faults or the breaking zone during or after coal mining, and therefore, the groundwater in the aquifer is considered to be one of the main threats to the safety of coal mining.

A total of 43 samples were collected from eight coal mines in northern Anhui Province, China. Concentrations of eight kinds of major ions (Na⁺, K⁺, Ca²⁺, Mg²⁺, Cl⁻, ⁠, ⁠, and ⁠) and total dissolved solids (TDS) were analyzed, and because of the low concentrations of K⁺, Na⁺, and K⁺ were merged to be one Na⁺ + K⁺. The analytical methods are as follows: Na⁺ + K⁺, Ca²⁺, Mg²⁺, Cl⁻, and were analyzed by ion chromatography, whereas and were analyzed by acid–base titration in the Engineering and Technological Research Center of Coal Exploration, Anhui Province, China.

All of the analytical results were first processed by Mystat software (version 12) and the min, max, median, mean, coefficient of variation, and p-value of Anderson–Darling test obtained. Then, the data were analyzed by EPA Unmix model (version 6) with the following processes: first, data pretreatment, concentrations were removed because most of the samples had concentrations equal to zero; second, factor analysis was processed to obtain the number of potential sources of major ions; third, all of the data were analyzed by Unmix model to get the number of sources, the compositions of the sources, and their contributions for each sample; fourth, the wall rock compositions and variations of the loose layer aquifer system were discussed according to the results obtained by Unmix model.

All of the analytical results are shown in Table 1. As can be seen from the table, these groundwater samples have the highest mean concentrations of Na⁺ + K⁺, and among other anions and cations, respectively. These groundwater samples can be classified to be Na-Cl type according to their concentrations. The mean concentrations of Na⁺ + K⁺, Ca²⁺, Mg²⁺, Cl⁻, ⁠, ⁠, and are 303 mg/L, 112 mg/L, 66.3 mg/L, 322 mg/L, 377 mg/L, 438 mg/L, and 8.20 mg/L, respectively.

A previous study (Davis & De Wiest 1966) classified groundwater on the basis of TDS into three kinds: <500 mg/L means desirable for drinking, 500–1,000 mg/L means permissible for drinking, and >3,000 mg/L can only be used for agricultural purposes. Based on this classification, the groundwater samples in this study must be treated before drinking.

Moreover, the coefficient of variation is always used for identifying anthropogenic contributions for the concentrations of pollutants in environmental studies: a low coefficient of variation (<10%) indicates a low degree of anthropogenic contribution, whereas a high coefficient of variation (>90%) indicates high degrees of anthropogenic contribution (Zhang & McGrath 2004). In this study, major ion concentrations of the groundwater samples have coefficients of variations ranging from 0.41 to 2.49, which indicates that all of them are statistically inhomogeneous and cannot be generated by a single source. This consideration is also supported by their p-values of the Anderson–Darling test because all of the major ions have p-values lower than 0.05 except for Na⁺ + K⁺, and ⁠.

Factor analysis has long been used for identifying the source of pollutants (Garcia et al. 1996; Lin et al. 2002; Liu et al. 2003) because of its easy processing. In this study, three factors have been obtained with eigenvalue higher than one after varimax rotation (Table 2), and the total variance explanation is 92.3%. The first factor, which accounts for 54.0% information, is dominated by Ca²⁺, Mg²⁺, ⁠, and TDS; factor 2 accounts for 22.1% information and is dominated by Cl⁻ and ⁠, whereas factor 3 accounts for 16.2% information and is dominated by Na⁺ + K⁺. These results suggest that at least three sources are responsible for the major ion concentrations in this study. However, factor analysis can only give the number of potential sources, and another method is needed for calculating the contributions of sources.

Based on the calculation of Unmix model, three sources have been identified and the results are listed in Table 3. These three sources have Min Rsq = 0.96, indicating that more than 96% of the variance information can be explained by the modeling and it is higher than the minimum requirement of the model (Min Rsq > 0.8). Moreover, the Min Sig/Noise is 3.14, also higher than the minimum requirement (Min Sig/Noise > 2). It can also be obtained from Figure 2 that the relationship between predicted and observed values of TDS is significant (r² = 0.98), suggesting that the modeling is efficient (Ai et al. 2014). The detailed explanations about these three sources are as follows.

Source 1 has the highest loadings of Cl⁻, and moderate loadings of Na⁺ + K⁺ among the three sources. This source has 69% and 27% contributions for Cl⁻ and Na⁺ + K⁺, respectively (Table 3). This source can be explained as the chloride source, such as halite (NaCl) in the strata. The contribution of the source for the TDS is 10%.

Source 2 has highest loadings of and Na⁺ + K⁺, and moderate loading of Ca²⁺ and Mg²⁺ among the three sources; 84%, 62%, 29%, and 20% of the ⁠, Na⁺ + K⁺, Mg²⁺, and Ca²⁺ are contributed by this source, respectively. This source can be explained as the silicate source, and to a lesser extent, carbonate source. The representative minerals are plagioclase and calcite, because the weathering of plagioclase can release Na⁺ and into the groundwater simultaneously, whereas dissolution of calcite can release Ca²⁺ and into the groundwater simultaneously. The contribution of the source for the TDS is 47%.

Source 3 has the highest loadings of Ca²⁺, Mg²⁺, and ⁠. The contributions of this source for Ca²⁺, Mg²⁺, and are 80%, 71%, and 96%, respectively. This source can be explained as the sulfate source, such as gypsum and mirabilite. The contribution of the source for the TDS is 43%.

The compositions of underground rocks are still something of a mystery for all of us. Although we can get information based on drilling cores, which can be obtained from drilling programs, there are still some unknown areas because drilling can only obtain the information from one point. However, the underground water, which has flowed over large areas, should have more information about the wall rocks because the chemical compositions of the groundwater are controlled water–rock interactions. Therefore, the chemical compositions of groundwater can be used for the inversion of wall rock compositions, at least the minerals incorporated in the water–rock interactions (Sun et al. 2011).

As can be seen from Figure 3, the contributions of each source for the groundwater samples in this study vary significantly, which might be an indication that the mineral compositions of the loose layer aquifer system are inhomogeneous. All of the samples can be subdivided into two categories based on the contributions from sources 1 to 3 (Figure 3).

First, most of the samples (32) have low contributions (<10%) from source 1, which indicates that the concentrations of chloride minerals (e.g., halite) in most of the areas in the loose layer aquifer system are low. However, 11 samples (6, 7, 10, 12–19) have high contributions from source 1. Second, 29 samples have contributions from source 2 lower than 25%, whereas 14 samples have contributions more than 25%, and suggest that the contributions from the weathering of silicate minerals or dissolution of carbonate minerals changed from area to area. Third, 17 samples have high contributions (>10%) from source 3, whereas other samples (26) have very low contributions from source 3 (some of them have contributions near zero).

Other interesting information obtained from Figure 3 is that most of the samples with low contributions from source 2 have high contributions from sources 1 and 3, whereas the samples with high contributions from source 2 have low contributions from sources 1 and 3. In consideration with the different means of these three sources, along with the different abilities of weathering resistance of the minerals (silicate > carbonate > sulfate > chloride), this phenomenon can be explained to be the result of: (1) the mineral compositions in the loose layer aquifer system are inhomogeneous and (2) the samples with high contributions from sources 1 and 3 are located in the discharge zone, whereas the samples with high contributions from source 2 are located in the recharge zone.

Based on the analysis of major ion concentrations from the loose layer aquifer system in northern Anhui Province, China by Unmix model, the following conclusions have been obtained:

• (1) The groundwater samples are classified to be Na-Cl type according to their major ion concentrations, and they cannot be used for drinking directly according to their high TDS contents.

• (2) Coefficients of variations, as well as the p-values of the Anderson–Darling test of the major ion concentrations suggest that more than one source is responsible for the major ion concentrations.

• (3) Three sources have been identified by Umix model, including chloride, silicate-carbonate, and sulfate sources.

• (4) The contributions of the three sources for the samples vary significantly, which can be explained by the inhomogeneous mineral compositions and the different locations (recharge or discharge) from which the samples were collected.

This work was financially supported by the National Natural Science Foundation of China (41302274), the Foundation of Scholarship Leaders in Suzhou University (2014XJXS05), and the Foundation of Scientific Platform in Suzhou University (2014YKF05).