Abstract

Nen River flows through Inner Mongolia, Heilongjiang and Jilin Provinces, which form the neighboring sections at the border of the provinces. In this study, 10 water parameters covering 11 sites were detected from 2011 to 2015. Cluster analysis (CA) was used to assess the spatial variation of Nen River, and grouped 11 sampling sites into three clusters (low, moderate and high pollution). Nen River was more highly influenced by the tributary of Yin River. Along the mainstream of Nen River, the water quality downstream was worse than upriver. The major pollution sources were agricultural activity, industrial drainage and city populations. With government management, the point source pollution upstream was well controlled, and the water quality of Nen River improved obviously in 2015. Grey relational analysis (GRA) combined with principal component analysis (PCA) and pairwise correlation analysis showed NH4+-N and total phosphate (TP) were principal factors influencing the water quality of Nen River. The TP in most sites of Nen River was still in high concentration (higher than 0.1 mg L–1). It was mainly caused by nonpoint source pollution that should be effectively measured in the future. This is the first time the water quality of the neighboring sections along Nen River has been assessed.

INTRODUCTION

Surface waters are easily accessible for the disposal of wastewater and run-off from agricultural land, thus they are most vulnerable to pollution. In many countries, especially in developing countries, water quality deterioration has become a critical issue that restricts socio-economic development (Liu et al. 2015). In recent years, water resource assessment and management were highly valued and widely studied (Liang et al. 2015; Scott & Chafer 2018).

Nen River is located in the northeast of China and is 1,370 km in length. It flows from the north of Heilongjiang province and Inner Mongolia to the south of Heilongjiang Province, and merges with Songhua River near Da'an at the neighboring region between Heilongjiang and Jilin Province. Nen River is the longest tributary of the Songhua River which is one of the primary rivers of China. The water quality of Nen River affects Songhua River directly. Nen River flows through Inner Mongolia, Heilongjiang and Jilin Province, forming the neighboring sections at the border of the provinces, which makes water management and policy decisions more complicated. Thus, the water quality of Nen River, especially of the neighboring sections, has not raised concern for decades. In 2008, the Law of Water Pollution Control in China was amended and the Water Resource Protection Program was developed. Under this program, water quality investigation of the neighboring sections along Nen River began in 2011. The water parameters' data sets were collected.

Multivariate statistical techniques, such as factor analysis (FA), principal component analysis (PCA), cluster analysis (CA), Grey relational analysis (GRA), etc., have been widely used in water quality assessment (Liu et al. 2009; Schaefer & Einax 2010; Mustapha & Aris 2012; Du et al. 2017). Hierarchical agglomerative CA is a method applied in grouping a matrix of data into classes based on their similarities within a class and dissimilarities between different classes (Jiang et al. 2015; Hajigholizadeh & Melesse 2017; Lifshitz & Avi Ostfeld 2018). PCA is the most powerful pattern recognition technique used, coupled with the CA method in water quality assessment. It is used in describing the dispersion of the multiple measured parameters to obtain eigenvalues and eigenvectors (Ogwueleka 2015). PCA groups variables in different varifactors, which helps in data interpretation. GRA can be used to investigate the basic relationship between influencing factors and the selected object, and determine the key influencing factor (Liu et al. 2009). This method has been used in evaluating water hydrology (Wang 2014; Qiu et al. 2017) and assessing the water pollution risk (Chen et al. 2007; Li et al. 2015). Normally, a combination of several methods is required for assessing water quality.

Previously, the water management of Nen River was mainly focused on the water quantity (Li et al. 2014; Wang et al. 2015). There was a lack of knowledge about the water quality of Nen River. Thus, it was difficult to determine the water pollution source and make policy decisions. In this study, a data matrix was obtained from 11 neighboring sites along Nen River over five years (2011–2015) for 10 factors including temperature, pH, dissolved oxygen (DO), total organic carbon (TOC), chemical oxygen demand (COD), biological oxygen demand-5 days (BOD5), NH4+-N, total phosphate (TP), arsenic and fluoride. CA, GRA, PCA and pairwise correlation analysis were introduced to: (1) evaluate the spatial variation of the water quality of Nen River; (2) classify the water quality of 11 neighboring sections along Nen River; (3) identify the key factor influencing the water quality of Nen River and determine the pollution source. This is the first time the water quality of Nen River has been assessed, including the neighboring sections between Heilonjiang, Jilin and Inner Mongolia. The results obtained in this study will be valuable for provincial government in water pollution control and water environment management.

METHODS

Study area

In this study, 11 sampling sites located along the main stream of Nen River and its tributaries between Heilonjiang, Jilin and Inner Mongolia were labeled as JXC, SHY, FRX, XMD, EWK, DH, JSW, MH, LJZ, DA and MKT. The sites of JXC (50°24′38″N, 124°5′2″E) and SHY (50°2′55″N, 125°19′22″E) are located at the source of Nen River. The site of FRX (49°7′24″N, 125°6′20″E) is near to the Nierji reservoir and the county of Nenjiang. The site of XMD (48°24′7″N, 124°30′49″E) is at the end of Nierji reservoir. The water quality of the site of XMD can reflect the influence of Nierji reservoir on Nen River's quality. The site of EWK (48°7′41″N, 124°30′18″E) is near to the county of Nehe, where a sugar-producing factory is located. The sites of DH (47°56′34″N, 123°22′30″E), JSW (47°39′26″N, 122°51′25″E) and MH (46°59′27″N, 123°32′58″E) are located near to the city of Qiqihar, which is the second largest city (42,205.82 km2 with seven districts, eight counties and one county-level city) in Heilongjiang Province. The sites of DA (45°32′5″N, 124°17′7″E) and MKT (45°26′51″N, 124°35′59″E) are near to Songhua River. The sites of JXC, DH, JSW and LJZ (46°44′20″N, 122°59′19″E) are located in the Nen River's tributaries, Gan River, Yin River, Yalu River and Chuoer River, respectively. Water flows from Inner Mongolia through the sites of JXC, DH, JSW and LJZ into Heilongjiang Province (from west to east). The site of DA is located at the end of Chaor River which is a tributary of Nen River and flows from Jilin Province through Da'an city into Heilongjiang Province. The sites of SHY, FRX, XMD, EWK and MH are located in the main stream of Nen River. Figure 1 shows the sampling sites along Nen River.

Figure 1

The location of Nen River and water sampling sites along the Nen River.

Figure 1

The location of Nen River and water sampling sites along the Nen River.

Sampling and pretreatment

The sampling and pretreatment procedures used were according to Technical Regulations for Evaluation of Surface Water Quality (SL395-2007). The samples were taken once per month in each site. Five sampling points were chosen in each site. After sampling, the water from the five points was combined, kept at 4 °C and transferred to the lab immediately. For detecting DO, the DO bottles were used to which were immediately added fixing agent containing MnSO4, KI and NaN3.

Analytical methods

Six thousand six hundred pieces of data were collected from 11 sampling sites over five years (5 years × 12 months × 11 sampling sites × 10 parameters). The NH4+-N concentration was measured by ammonium-Nessler's reagent colorimetric method (Zheng et al. 2018). A 50.0 mL water sample was mixed with 1.0 mL sodium potassium tartrate (50%, w/w) and Nessler's reagent (7% KI, 10% HgI2, and 16% NaOH, w/w). After a 10 min reaction, the mixture was detected at a wavelength of 420 nm with a spectrophotometer 721G-100 (INESA, China).

COD was detected using a chemical analysis method (Ministry of Environmental Protection of the People's Republic of China 2002). A 3.0 mL water sample was mixed with 1.0 mL masking agent (10 g HgSO4 dissolved in 100 mL H2SO4), 3.0 mL digestion solution (0.4 mol L–1 K2Cr2O7) and 5.0 mL catalyzer (8.8 g Ag2SO4 dissolved in 1,000 mL H2SO4). The mixture was sealed and heated at 165 °C for 10 min, then was detected at a wavelength of 600 nm with a spectrophotometer 721G-100 (INESA, China).

Fluoride and TP were detected with Ion Chromatography ICS-1100 (Dionex, USA). TOC was determined by a TOC analyzer TOC-4200 (Shimadzu, Japan). Temperature and pH were detected with HI9829 (HANNA, Italy). The DO was detected with HI9146 (HANNA, Italy). As was determined by an ICP-AES Optima 7000DV (PerkinElmer, USA). The average data of each sampling site are shown in Table 1.

Table 1

Summary of basic statistics of water quality parameters in each sampling site based on monthly averages of variable (from 2011 to 2015)

Site number Statistics Temperature (°C) pH DO (mg L–1TOC (mg L–1COD (mg L–1BOD5 (mg L–1NH4+-N (mg L–1TP (mg L–1As (mg L–1Fluoride (mg L–1
JXC Min 4.90 7.38 7.70 2.93 12.05 1.62 0.19 0.02 0.0016 0.13 
Max 6.20 7.50 8.81 4.70 15.60 1.91 0.30 0.15 0.0035 0.38 
Mean 5.67 7.47 8.33 3.69 14.39 1.77 0.24 0.06 0.0026 0.27 
SD 0.42 0.048 0.40 0.63 1.22 0.10 0.04 0.05 0.0009 0.08 
SHY Min 7.30 6.86 8.03 4.65 16.09 1.46 0.14 0.03 0.0009 0.12 
Max 8.74 7.50 9.95 6.30 20.20 2.75 0.40 0.19 0.0026 0.24 
Mean 7.92 7.26 9.07 5.78 19.14 2.00 0.30 0.09 0.0016 0.17 
SD 0.51 0.21 0.62 0.60 1.57 0.44 0.09 0.05 0.0005 0.41 
FRX Min 8.00 7.39 8.37 4.71 17.29 2.15 0.39 0.10 0.0012 0.16 
Max 9.80 7.60 9.62 7.10 26.80 2.61 0.55 0.26 0.0019 0.24 
Mean 8.96 7.52 9.00 5.76 21.82 2.38 0.45 0.17 0.0017 0.18 
SD 0.57 0.10 0.48 0.80 3.64 0.19 0.06 0.06 0.0003 0.03 
XMD Min 8.10 7.30 8.83 4.48 15.81 2.11 0.34 0.07 0.0019 0.19 
Max 9.80 7.70 9.93 7.70 24.59 2.80 0.43 0.15 0.0026 0.23 
Mean 9.28 7.48 9.13 5.67 20.21 2.37 0.38 0.12 0.0023 0.22 
SD 0.62 0.13 0.41 1.09 2.87 0.28 0.04 0.03 0.0003 0.02 
EWK Min 7.40 7.60 7.76 4.81 20.16 2.00 0.29 0.04 0.0013 0.14 
Max 9.96 7.80 8.95 6.25 13.51 2.25 0.60 0.18 0.0040 0.33 
Mean 9.15 7.69 8.44 5.42 17.67 2.13 0.40 0.12 0.0022 0.21 
SD 0.97 0.09 0.40 0.82 2.36 0.81 0.11 0.06 0.0009 0.07 
DH Min 8.40 7.50 8.15 2.38 11.92 1.45 0.19 0.03 0.0019 0.16 
Max 10.10 7.60 10.10 5.46 21.76 3.15 0.74 0.28 0.0039 0.48 
Mean 9.52 7.52 8.94 3.70 16.24 2.31 0.39 0.13 0.0027 0.28 
SD 0.59 0.04 0.69 1.11 3.90 0.69 0.19 0.10 0.0007 0.11 
JSW Min 8.70 7.30 8.25 2.41 10.98 1.72 0.36 0.10 0.0020 0.21 
Max 10.30 7.70 10.21 3.72 17.12 2.25 0.66 0.18 0.0034 0.33 
Mean 9.66 7.52 8.91 3.01 14.03 2.03 0.47 0.14 0.0027 0.28 
SD 0.56 0.15 0.72 0.44 2.06 0.20 0.11 0.03 0.0005 0.04 
MH Min 8.57 7.62 7.70 4.23 14.88 1.80 0.28 0.04 0.0021 0.20 
Max 10.10 7.90 8.88 5.50 21.08 2.78 0.63 0.19 0.0026 0.26 
Mean 9.23 7.76 8.55 4.94 17.09 2.17 0.50 0.14 0.0023 0.23 
SD 0.59 0.10 0.44 0.50 2.23 0.35 0.16 0.06 0.0002 0.02 
LJZ Min 9.30 7.90 7.94 2.28 10.00 1.70 0.16 0.01 0.0029 0.29 
Max 10.90 8.10 8.59 3.03 12.48 3.55 0.36 0.12 0.0033 0.32 
Mean 10.16 7.94 8.33 2.71 11.66 2.30 0.22 0.06 0.0031 0.31 
SD 0.64 0.08 0.23 0.27 0.88 0.70 0.07 0.04 0.0001 0.01 
DA Min 9.50 7.70 7.91 4.31 15.88 2.47 0.42 0.12 0.0009 0.10 
Max 11.70 7.60 9.83 5.71 20.10 3.32 0.63 0.20 0.0030 0.40 
Mean 10.44 7.68 8.79 4.97 18.51 2.77 0.54 0.17 0.0024 0.24 
SD 0.93 0.04 0.73 0.53 1.42 0.29 0.08 0.03 0.0013 0.11 
MKT Min 9.50 8.10 7.79 3.92 15.29 2.16 0.48 0.14 0.0017 0.18 
Max 11.80 7.60 9.77 5.11 18.94 3.25 0.96 0.30 0.0034 0.30 
Mean 10.46 7.70 8.76 4.49 16.77 2.57 0.67 0.21 0.0025 0.24 
SD 0.87 0.20 0.63 0.43 1.37 0.37 0.17 0.05 0.0006 0.04 
Site number Statistics Temperature (°C) pH DO (mg L–1TOC (mg L–1COD (mg L–1BOD5 (mg L–1NH4+-N (mg L–1TP (mg L–1As (mg L–1Fluoride (mg L–1
JXC Min 4.90 7.38 7.70 2.93 12.05 1.62 0.19 0.02 0.0016 0.13 
Max 6.20 7.50 8.81 4.70 15.60 1.91 0.30 0.15 0.0035 0.38 
Mean 5.67 7.47 8.33 3.69 14.39 1.77 0.24 0.06 0.0026 0.27 
SD 0.42 0.048 0.40 0.63 1.22 0.10 0.04 0.05 0.0009 0.08 
SHY Min 7.30 6.86 8.03 4.65 16.09 1.46 0.14 0.03 0.0009 0.12 
Max 8.74 7.50 9.95 6.30 20.20 2.75 0.40 0.19 0.0026 0.24 
Mean 7.92 7.26 9.07 5.78 19.14 2.00 0.30 0.09 0.0016 0.17 
SD 0.51 0.21 0.62 0.60 1.57 0.44 0.09 0.05 0.0005 0.41 
FRX Min 8.00 7.39 8.37 4.71 17.29 2.15 0.39 0.10 0.0012 0.16 
Max 9.80 7.60 9.62 7.10 26.80 2.61 0.55 0.26 0.0019 0.24 
Mean 8.96 7.52 9.00 5.76 21.82 2.38 0.45 0.17 0.0017 0.18 
SD 0.57 0.10 0.48 0.80 3.64 0.19 0.06 0.06 0.0003 0.03 
XMD Min 8.10 7.30 8.83 4.48 15.81 2.11 0.34 0.07 0.0019 0.19 
Max 9.80 7.70 9.93 7.70 24.59 2.80 0.43 0.15 0.0026 0.23 
Mean 9.28 7.48 9.13 5.67 20.21 2.37 0.38 0.12 0.0023 0.22 
SD 0.62 0.13 0.41 1.09 2.87 0.28 0.04 0.03 0.0003 0.02 
EWK Min 7.40 7.60 7.76 4.81 20.16 2.00 0.29 0.04 0.0013 0.14 
Max 9.96 7.80 8.95 6.25 13.51 2.25 0.60 0.18 0.0040 0.33 
Mean 9.15 7.69 8.44 5.42 17.67 2.13 0.40 0.12 0.0022 0.21 
SD 0.97 0.09 0.40 0.82 2.36 0.81 0.11 0.06 0.0009 0.07 
DH Min 8.40 7.50 8.15 2.38 11.92 1.45 0.19 0.03 0.0019 0.16 
Max 10.10 7.60 10.10 5.46 21.76 3.15 0.74 0.28 0.0039 0.48 
Mean 9.52 7.52 8.94 3.70 16.24 2.31 0.39 0.13 0.0027 0.28 
SD 0.59 0.04 0.69 1.11 3.90 0.69 0.19 0.10 0.0007 0.11 
JSW Min 8.70 7.30 8.25 2.41 10.98 1.72 0.36 0.10 0.0020 0.21 
Max 10.30 7.70 10.21 3.72 17.12 2.25 0.66 0.18 0.0034 0.33 
Mean 9.66 7.52 8.91 3.01 14.03 2.03 0.47 0.14 0.0027 0.28 
SD 0.56 0.15 0.72 0.44 2.06 0.20 0.11 0.03 0.0005 0.04 
MH Min 8.57 7.62 7.70 4.23 14.88 1.80 0.28 0.04 0.0021 0.20 
Max 10.10 7.90 8.88 5.50 21.08 2.78 0.63 0.19 0.0026 0.26 
Mean 9.23 7.76 8.55 4.94 17.09 2.17 0.50 0.14 0.0023 0.23 
SD 0.59 0.10 0.44 0.50 2.23 0.35 0.16 0.06 0.0002 0.02 
LJZ Min 9.30 7.90 7.94 2.28 10.00 1.70 0.16 0.01 0.0029 0.29 
Max 10.90 8.10 8.59 3.03 12.48 3.55 0.36 0.12 0.0033 0.32 
Mean 10.16 7.94 8.33 2.71 11.66 2.30 0.22 0.06 0.0031 0.31 
SD 0.64 0.08 0.23 0.27 0.88 0.70 0.07 0.04 0.0001 0.01 
DA Min 9.50 7.70 7.91 4.31 15.88 2.47 0.42 0.12 0.0009 0.10 
Max 11.70 7.60 9.83 5.71 20.10 3.32 0.63 0.20 0.0030 0.40 
Mean 10.44 7.68 8.79 4.97 18.51 2.77 0.54 0.17 0.0024 0.24 
SD 0.93 0.04 0.73 0.53 1.42 0.29 0.08 0.03 0.0013 0.11 
MKT Min 9.50 8.10 7.79 3.92 15.29 2.16 0.48 0.14 0.0017 0.18 
Max 11.80 7.60 9.77 5.11 18.94 3.25 0.96 0.30 0.0034 0.30 
Mean 10.46 7.70 8.76 4.49 16.77 2.57 0.67 0.21 0.0025 0.24 
SD 0.87 0.20 0.63 0.43 1.37 0.37 0.17 0.05 0.0006 0.04 

Data preparation

For CA, to avoid classification problems with objects described by variables of completely different size, the data set was auto-scaled with Equations (1) and (2) (Schaefer & Einax 2010): 
formula
(1)
where xij is the data of each parameter in each sampling site; is the average of each parameter; Sj is the standard deviation of each parameter: 
formula
(2)
For GRA analysis, the data were treated as none-dimensional data with Equations (3) and (4): 
formula
(3)
 
formula
(4)

Multivariate statistical methods

Hierarchical agglomerative CA was used for clustering water sampling sites and classifying the water quality of each site. For hierarchical agglomerative CA, the squared Euclidean distances were selected as a measure of similarity because it decreased the importance of small distances (Hajigholizadeh & Melesse 2017). The Ward's method (Willett 1987), which was selected in this study, was the most commonly used method for grouping the cases (Shrestha & Kazama 2007; Guo et al. 2012; Taoufik et al. 2017). The results of hierarchical agglomerative CA were presented by the linkage distance, reported as Dlink/Dmax, which represented the quotient between the linkage distances for a particular case divided by the maximal linkage distance. The quotient was then multiplied by 100 as a way to standardize the linkage distance represented on the y-axis (Shrestha & Kazama 2007).

GRA was used to analyze the correlation of temperature, pH, DO, TOC, COD, BOD5, NH4+-N, TP, arsenic and fluoride with the water quality of Nen River. The data were prepared according to Equations (3) and (4) (Liu et al. 2009). The coefficient is calculated by using Equation (5): 
formula
(5)
where ζ is the recognition differential, a value of 0.5 was used in this paper. 
formula
(6)
The correlation degree Ri was calculated with Equation (7): 
formula
(7)
where m is the number of data set of each factor.

PCA of the data set was performed to extract significant PCs and to further reduce the contribution of variables with minor significance. The principle factors influencing the water quality were subject to pairwise correlation analysis in this study. Multivariate analysis in this study was conducted with SPSS 19.0. The maps were draw with MapGIS 10.2.

RESULTS AND DISCUSSION

Water quality assessment of Nen River

The dendrogram developed by CA with SPSS 19.0 is shown in Figure 2. The linkage distance, shown by Dlink/Dmax, was used to evaluate the variable degree between each sampling site. Eleven sampling sites were clustered into three groups, which were low polluted (LP), moderate polluted (MP) and high polluted (HP). Figure 3 demonstrates the spatial distribution of three groups. The sites of JXC, JSW and LJZ were clustered in one group, which showed good water quality with low pollution; however, the sites of JXC and SHY located at the original region of Nen River showed different water quality. The water quality of the SHY site was worse than the JXC site. The reason for this may be that the site of SHY was surrounded with large areas of farmland (Jianbian Farm with 188 km2 farmland) with higher agricultural activity than the site of JXC. The site of JSW is located in the west of the mountain region (Nianzi Mountain) which is about 80 km to the west of Qiqihaer's main urban area. On the east side of JSW, from north to south, there is a large area of forest with fewer human activities. The site of LJZ is surrounded with a large region of wet land in a good ecological environment, which is why the sites of JSW and LJZ showed good water quality, although they were downstream of Nen River. The sites of FYX and XMD located in the middle of Nen River showed moderate water quality. The sites of EWK, DH, MH, DA and MKT located downstream of Nen River showed poor water quality. Three groups of sampling sites were generated in a convincing way by CA.

Figure 2

Dendrogram developed by cluster analysis with SPSS 19.0. It shows three clusters of 11 sampling sites.

Figure 2

Dendrogram developed by cluster analysis with SPSS 19.0. It shows three clusters of 11 sampling sites.

Figure 3

Spatial distribution of three groups of sampling sites based on the results of cluster analysis.

Figure 3

Spatial distribution of three groups of sampling sites based on the results of cluster analysis.

Key factors influencing the water quality

In this study, the water quality of Nen River was seen as a grey problem. The GRA method was used to demonstrate the relationship between water parameters and water quality. The results are shown in Table 2. In general, when GRA correlation degree Ri is above 0.9, it means a marked effect; when Ri is between 0.8 and 0.9, it means a relatively marked effect; when Ri is between 0.7 and 0.8, it means a noticeable effect; when Ri is lower than 0.6, it means a negligible effect (Liu et al. 2009). In this study, pH showed relatively marked correlation with water quality (Ri was 0.83). Temperature and BOD5 showed a noticeable correlation with water quality (Ri was 0.74 and 0.71, respectively). NH4+-N and TP also showed higher correlation with water quality (Ri were 0.69 and 0.64, respectively). DO showed the lowest correlation with water quality (Ri was 0.54). These results were obtained based on the data of all sampling sites over five years. To prove the reliability of the results, we calculated the grey correlation degrees between water parameters and each sampling site. The results are shown in Table 3. Unfortunately, the grey correlation degrees presented in an irregular way. It is difficult to determine the main factors influencing the water quality of each sampling site. Thus, PCA was introduced to analyze the main factors influencing the water quality of Nen River in this study.

Table 2

Correlation degree (Ri) between water quality of Nen River and water parameters (DO, TOC, COD, BOD5, NH4+-N, TP, arsenic and fluoride) based on all databases

Temperature (°C) pH DO TOC COD BOD5 NH4+-N TP As Fluoride 
0.74 0.83 0.54 0.63 0.58 0.71 0.69 0.64 0.66 0.60 
Temperature (°C) pH DO TOC COD BOD5 NH4+-N TP As Fluoride 
0.74 0.83 0.54 0.63 0.58 0.71 0.69 0.64 0.66 0.60 
Table 3

Correlation degree (Ri) between the water quality of each sampling site and water parameters including DO, COD, TOC, BOD5, NH4+-N, TP, arsenic and fluoride

Site number Temperature (°C) pH DO TOC COD BOD5 NH4+-N TP As Fluoride 
JXC 0.69 0.90 0.60 0.60 0.73 0.70 0.56 0.82 0.81 0.82 
SHY 0.67 0.69 0.65 0.66 0.66 0.62 0.65 0.62 0.78 0.69 
FRX 0.71 0.91 0.69 0.77 0.73 0.71 0.70 0.79 0.67 0.76 
XMD 0.68 0.71 0.65 0.68 0.68 0.85 0.73 0.88 0.53 0.72 
EWK 0.77 0.69 0.77 0.76 0.78 0.62 0.69 0.79 0.76 0.71 
DH 0.65 0.90 0.80 0.77 0.54 0.50 0.68 0.67 0.68 0.68 
JSW 0.80 0.81 0.76 0.68 0.58 0.58 0.67 0.61 0.60 0.76 
MH 0.57 0.76 0.66 0.72 0.70 0.65 0.60 0.87 0.89 0.75 
LJZ 0.77 0.90 0.74 0.65 0.66 0.77 0.76 0.81 0.53 0.78 
DA 0.73 0.90 0.54 0.52 0.66 0.67 0.68 0.64 0.60 0.66 
MKT 0.68 0.90 0.64 0.54 0.65 0.67 0.69 0.71 0.67 0.55 
Site number Temperature (°C) pH DO TOC COD BOD5 NH4+-N TP As Fluoride 
JXC 0.69 0.90 0.60 0.60 0.73 0.70 0.56 0.82 0.81 0.82 
SHY 0.67 0.69 0.65 0.66 0.66 0.62 0.65 0.62 0.78 0.69 
FRX 0.71 0.91 0.69 0.77 0.73 0.71 0.70 0.79 0.67 0.76 
XMD 0.68 0.71 0.65 0.68 0.68 0.85 0.73 0.88 0.53 0.72 
EWK 0.77 0.69 0.77 0.76 0.78 0.62 0.69 0.79 0.76 0.71 
DH 0.65 0.90 0.80 0.77 0.54 0.50 0.68 0.67 0.68 0.68 
JSW 0.80 0.81 0.76 0.68 0.58 0.58 0.67 0.61 0.60 0.76 
MH 0.57 0.76 0.66 0.72 0.70 0.65 0.60 0.87 0.89 0.75 
LJZ 0.77 0.90 0.74 0.65 0.66 0.77 0.76 0.81 0.53 0.78 
DA 0.73 0.90 0.54 0.52 0.66 0.67 0.68 0.64 0.60 0.66 
MKT 0.68 0.90 0.64 0.54 0.65 0.67 0.69 0.71 0.67 0.55 

According to the results obtained from CA, the water quality of each sampling site was set as 1 (low polluted), 2 (moderate polluted) and 3 (high polluted). The water quality of Nen River was set as one of the factors, calculated with water parameters data set by the PCA method using SPSS 19.0. The results are shown in Figure 4. PC1 and PC2 explained 78.43 and 16.56% of total variance, respectively. PC1 and PC2 showed strong positive loadings on BOD5, NH4+-N and TP, which were grouped with the water quality of Nen River. For further confirmation of the correlation degree between the water parameters and water quality of Nen River, pairwise correlation analysis was conducted. The results are shown in Table 4. Water parameters correlated with each other. pH and As, DO and COD, BOD5 and NH4+-N, BOD5 and TP, showed significantly pairwise correlation (P ≤ 0.05 level) with each other. Temperature and BOD5, TOC and COD, NH4+-N and TP, As and fluoride showed significant pairwise correlation (P ≤ 0.01 level) with each other. In all water parameters, NH4+-N and TP showed significant correlation (P ≤ 0.05 level) with the water quality of Nen River.

Table 4

The Pearson correlation results after pairwise correlation analysis for water parameters and water quality

Water parameters Temperature (°C) pH DO TOC COD BOD5 NH4+-N TP As Fluoride 
pH 0.561          
DO 0.212 −0.648**         
TOC −0.070 −0.394 0.461        
COD 0.011 −0.469 0.659* 0.934**       
BOD5 0.798** 0.401 0.285 0.211 0.331      
NH4+-N 0.589 0.175 0.291 0.246 0.333 0.628*     
TP 0.575 0.080 0.419 0.329 0.475 0.680* 0.966**    
As 0.252 0.602* −0.515 −0.888** −0.857** 0.044 −0.124 −0.235   
Fluoride 0.162 0.507 −0.484 −0.948** −0.885** −0.046 −0.199 −0.292 0.980**  
Water quality 0.425 0.145 0.160 0.510 0.473 0.552 0.639* 0.626* −0.274 −0.364 
Water parameters Temperature (°C) pH DO TOC COD BOD5 NH4+-N TP As Fluoride 
pH 0.561          
DO 0.212 −0.648**         
TOC −0.070 −0.394 0.461        
COD 0.011 −0.469 0.659* 0.934**       
BOD5 0.798** 0.401 0.285 0.211 0.331      
NH4+-N 0.589 0.175 0.291 0.246 0.333 0.628*     
TP 0.575 0.080 0.419 0.329 0.475 0.680* 0.966**    
As 0.252 0.602* −0.515 −0.888** −0.857** 0.044 −0.124 −0.235   
Fluoride 0.162 0.507 −0.484 −0.948** −0.885** −0.046 −0.199 −0.292 0.980**  
Water quality 0.425 0.145 0.160 0.510 0.473 0.552 0.639* 0.626* −0.274 −0.364 

*Significant correlation (P ≤ 0.05); **significant correlation (P ≤ 0.01).

Figure 4

The principal factors influencing the water quality results from PCA.

Figure 4

The principal factors influencing the water quality results from PCA.

In this study, the results of GRA, PCA and pairwise correlation analysis were not totally in accordance with each other, especially for the results of temperature and pH. GRA results showed the highest correlation between pH and water quality (Ri was 0.83), but PCA results showed that pH was negatively loading with PC1 (–0.447). Pearson correlation between pH and water quality was only 0.145. Both GRA and PCA results showed higher correlation between temperature and water quality, but Pearson correlation between temperature and water quality was low. Researchers usually used GRA or PCA/factor analysis to assess water quality (Juahir et al. 2011; Li et al. 2015; Ogwueleka 2015). In this study, GRA, PCA and factor analysis were all used to evaluate Nen River, but the results presented deviation. It is necessary to combine these results to deduce the conclusion.

Normally, seasonal variations and effects were analyzed by assessing the water quality (Hajigholizadeh & Melesse 2017; Basatnia et al. 2018). Unfortunately, we could only access the annual average data in this study. We will keep communicating with our partner to support monthly or seasonal data, and study the seasonal effects in the future.

From 2011 to 2015, the water quality of Nen River improved (annual data are not shown in this study). BOD5 of the sites DH, LJZ, DA and MKT exceeded 3.0 mg L–1 in 2011 and 2012, but decreased to 1.45–2.65 mg L–1 during 2013–2015. According to Environmental Quality Standards for Surface Water of P. R. China (GB3838-2002), when BOD5 was lower than 3.0 mg L–1, the water quality could be grouped as Class I and II (potable water resource with good quality). In 2015, BOD5 of all sampling sites of Nen River were lower than 3.0 mg L–1, which indicated that BOD5 was not the major managing factor for Nen River.

From 2011 to 2015, the TP of Nen River was well managed and decreased from 0.2–0.3 to 0.03–0.14 mg L–1. NH4+-N pollution also showed improvement. For example, the NH4+-N concentration of the site of DH decreased from 0.74 to 0.19 mg L–1. However, the NH4+-N (about 0.5 mg L–1) and TP (higher than 0.1 mg L–1) were still in high levels downstream of Nen River, especially for the sites of MH, DA and MKT. According to Environmental Quality Standards for Surface Water of P. R. China (GB3838-2002), when NH4+-N was 0.50–1.00 mg L–1, TP was lower than 0.10–0.20 mg L–1; the water quality could be grouped as Class III (potable water resource, but quality should be improved). Above all, NH4+-N and TP were the priority factors influencing water quality of Nen River, which should be well managed. The results are valuable for our future modeling work. In this study, a combination of GRA, PCA and pairwise correlation analysis was a conclusive way to assess the water quality of Nen River.

Pollution resource of Nen River

In general, Nen River showed better quality than most rivers in China (Zou et al. 2015; Wu et al. 2018). Especially from 2011, 13 typical sewage discharge points upstream of Nen River were investigated and well controlled, and the water quality of Nen River improved obviously. However, NH4+-N and TP pollution were still an issue, which were also typical pollution factors for rivers in China (Wu et al. 2018). The presence of nitrogen and TP in the river were mainly due to soil erosion and agricultural runoff (Juahir et al. 2011; Yue et al. 2014). Nen River is the largest tributary of Songhua River, which is one of the largest river basins in China. Nen River, especially its downstream, flows through the Songnen Plain with large areas of agricultural land (103,200 km2) (Yue et al. 2014), which would be the reason for the presentation of higher nitrogen and TP in this study.

In this study, we evaluated the water quality of the tributaries and main stream of Nen River. The results showed that the water quality of the tributaries was better than the main stream at similar latitude (shown in Figure 3). The site of JXC is located in the tributary of Gan River. The sites of JSW and LJZ are located in the tributaries of Yalu River and Chuoer River, respectively. Water flows through the sites of JXC, JSW and LJZ from Inner Mongolia to Heilongjiang province (from west to east). Regions around the sites of JXC, JSW and LJZ are covered with forest or grassland, which leads to less soil erosion and causes better water quality (Yue et al. 2014).

For the main stream of Nen River, water quality was worse downstream than upstream. The site of SHY is near to the source of Nen River. It was influenced by Jianbian Farm (with 188 km2 farmland) and showed worse water quality than the site of JXC. The site of FRX is located at the beginning of Nierji Reservoir and the site of XMD is located at the end of Nierji Reservoir. Water quality showed less difference between the site of FRX and XMD. Nierji Reservoir did not make an obvious impact on the water quality of Nen River. The pollution of the site of MH was mainly from upriver and the city of Qiqihar in Heilongjiang Province. There was a large-scale sugar producing factory near the site of EWK and it had a great impact on the water quality of Nen River because of its high strength wastewater. The site of DH is located in the Yin River, the tributary of Nen River. The water quality of Yin River had an impact on that of Nen River. The site of DA is located at the end of Chaor River which flows from Jilin to Heilongjiang Province. High pollution of the site of DA showed that the pollution source was mainly from Jinlin Province. Through the sites of MKT, water flows into Songhua River. The water quality of the site of MKT was at a high pollution level and influenced Songhua River directly; both Jilin and Heilongjiang Province were responsible for the lower water quality.

According to the results obtained in this study, preventive measures for the future are proposed as follows. (1) It is quite necessary to continue investigating and controlling the sewage discharge points, particularly downstream of Nen River. (2) Eutrophic parameters of Nen River, including the seasonal dynamics and correlation between alga, N and P, should be investigated, because the TP and NH4+-N in most of the sites were at high levels for long periods. (3) Nonpoint pollution source could be the key pollution source and should be well managed with collaboration between the departments of Environmental Protection, Agriculture, and Water Resources.

Practical application of this work

Multivariate statistical analysis has been widely used in water quality assessment. In this study, the results of CA, GRA, PCA and Pairwise Correlation Analysis were taken into account to deduce the major factors correlated with the water quality of Nen River. In this study, the water quality of the tributaries and main stream of Nen River were assessed for deciding the source of water pollution. This research can also be used in rivers located in neighboring regions, such as Amur River or Mekong River. For example, Mekong is a trans-boundary river in southeast Asia. It flows from the Tibetan Plateau, running through China, Myanmar, Laos, Thailand, Cambodia and Vietnam. Assessing the water quality of Mekong River's tributaries and mainstream, determining the major influencing factors and the pollution source will be of benefit for solving the water pollution issue with cooperation between China, Myanmar, Laos, Thailand, Cambodia and Vietnam.

CONCLUSIONS

In this study, 11 neighboring sections of Nen River were grouped into three clusters, low polluted (sites of JXC, JSW and LJZ), moderate polluted (sites of SHY, FRX and XMD) and high polluted (sites of EWK, DH, MH, DA and MKT). Nen River was less affected by tributaries of Gan River, Yalu River and Chuoer River, but was more influenced by the tributary of Yin River. Nierji reservoir did not show an obvious influence on the water quality of Nen River. The farmland of Jianbian and Songnen Plain, the Sugar factory and Qiqihar city impacted significantly on the water quality of Nen River. By GRA, PCA and Pairwise Correlation Analysis, NH4+-N and TP significantly correlated with the water quality of Nen River (Pearson coefficient was 0.639* and 0.626* respectively). The TP of most sites of Nen River were still in high concentration, mainly caused by nonpoint source pollution which should be effectively measured in the future.

ACKNOWLEDGEMENTS

This study was supported by grants from the National Natural Science Foundation of China (Grant No. 51608149), Harbin Application Technology Research and Development Project (2017RAQXJ072). We are grateful for the support from the Key Laboratory of Functional Inorganic Material Chemistry (Heilongjiang University) (Ministry of Education) and the Engineering Research Center of Rural Water Safety, Heilongjiang Province.

REFERENCES

REFERENCES
Basatnia
N.
,
Hossein
S. A.
,
Rodrigo-Comino
J.
,
Khaledian
Y.
,
Brevik
E. C.
,
Aitkenhead-Peterson
J.
&
Natesan
U.
2018
Assessment of temporal and spatial water quality in international Gomishan Lagoon, Iran, using multivariate analysis
.
Environ. Monit. Assess.
190
(
5
),
314
.
Chen
J.
,
Zhu
J. M.
,
Wang
Z. Y.
,
Liu
X. H.
&
Zhang
X. L.
2007
Application of grey relational analysis in water quality evaluation
.
J. Grey Syst.
19
(
1
),
99
106
.
Du
X. J.
,
Shao
F. J.
,
Wu
S. Y.
,
Zhang
H. L.
&
Xu
S.
2017
Water quality assessment with hierarchical cluster analysis based on Mahalanobis distance
.
Environ. Monit. Assess.
189
(
7
),
335
.
Juahir
H.
,
Zain
S. M.
,
Yusoff
M. K.
,
Hanidza
T. I. T.
,
Armi
A. S. M.
,
Toriman
M. E.
&
Mokhtar
M.
2011
Spatial water quality assessment of Langat River Basin (Malaysia) using environmentric techniques
.
Environ. Monit. Assess.
173
(
1–4
),
625
641
.
Li
H. Y.
,
Wang
Y. X.
,
Jia
L. N.
,
Wu
Y. N.
&
Xie
M.
2014
Runoff characteristics of the Nen River Basin and its cause
.
J. Mt. Sci.
11
(
1
),
110
118
.
Liang
J.
,
Yang
Q.
,
Sun
T.
,
Martin
J. D.
,
Sun
H.
&
Li
L.
2015
MIKE 11 model-based water quality model as a tool for the evaluation of water quality management plans
.
J. Water Supply Res. Technol. Aqua
64
(
6
),
708
718
.
Lifshitz
R.
&
Avi Ostfeld
F. A.
2018
Clustering for analysis of water distribution systems
.
J. Water Resour. Plan. Manage. ASCE
144
(
5
),
04018016-1-8
.
Liu
S. X.
,
Li
W.
,
Xie
Y. L.
,
Wang
B.
&
Huang
G. H.
2015
A coupled model for water resources allocation regarding water quality control
.
J. Water Supply Res. Technol. Aqua
64
(
8
),
870
882
.
Ministry of Environmental Protection of the People's Republic of China
2002
Water and Wastewater Analysis
,
4th edn
.
Environmental Science Press
,
Beijing
,
China
.
Mustapha
A.
&
Aris
A. Z.
2012
Spatial aspects of surface water quality in the Jakara Basin, Nigeria using chemometric analysis
.
J. Environ. Sci. Health Part A Toxic/Hazard. Subst. Environ. Eng.
47
(
10
),
1455
1465
.
Schaefer
K.
&
Einax
J. W.
2010
Analytical and chemometric characterization of the Cruces River in South Chile
.
Environ. Sci. Pollut. Res.
17
(
1
),
115
123
.
Scott
B.
&
Chafer
C. J.
2018
The pollution source assessment tool for Sydney's drinking water catchments
.
J. Water Supply Res. Technol. Aqua
67
(
1
),
44
53
.
Wang
J. F.
2014
Two-factor grey relational analysis method of the evaluation on fairness of two-dimensional water right allocation
.
J. Grey Syst.
26
(
3
),
23
31
.
Willett
J.
1987
Similarity and Clustering in Chemical Information Systems
.
John Wiley & Sons, Inc.
,
New Jersey
,
USA
.
Wu
Z.
,
Wang
X.
,
Chen
Y.
,
Cai
Y.
&
Deng
J.
2018
Assessing river water quality using water quality index in Lake Taihu Basin, China
.
Sci. Total Environ.
612
,
914
922
.
Yue
F. J.
,
Liu
C. Q.
,
Li
S. L.
,
Zhao
Z. Q.
,
Liu
X. L.
,
Ding
H.
,
Liu
B. J.
&
Zhong
J.
2014
Analysis of δ15n and δ18o to identify nitrate sources and transformations in Songhua River, Northeast China
.
J. Hydrol.
519
,
329
339
.
Zou
H.
,
Zou
Z.
&
Wang
X.
2015
An enhanced K-means algorithm for water quality analysis of the Haihe River in China
.
Int. J. Environ. Res. Public Health
12
,
14400
14413
.