International Journal of Environmental Science and Technology (2023) 20:2487–2502

[Link]

ORIGINAL PAPER

Water quality assessment of the Nam River, Korea, using multivariate

statistical analysis and WQI
H. G. Kwon1 · C. D. Jo1

Received: 27 January 2022 / Revised: 17 July 2022 / Accepted: 2 January 2023 / Published online: 7 January 2023
© The Author(s) 2023

Abstract
Water quality assessment using water quality index (WQI) is performed based on major variables reflecting the river charac‑
teristics. We evaluated the water quality of tributaries of the Nam River in South Korea. We analyzed the tributaries spatial
characteristics using cluster analyses and selected the main water quality variables (Cluster 1: chemical oxygen demand
[COD], total organic carbon [TOC], total nitrogen, and total phosphorus; Cluster 2: water temperature [WT], dissolved oxy‑
gen [DO], COD, and TOC; Cluster 3: WT, DO, electrical conductance, COD, and TOC) of the clustered rivers. The WQI
for each tributary was calculated using variables selected post statistical analysis. We verified that the WQI calculated in
this study was similar to the annual change in water quality of the target river. Finally, it was analyzed that performing river
water quality evaluation using the major variables selected using statistical analysis reflects the current water quality status
of the river in more detail. (WQI grade was S1 Good (63.0), S2 Poor (53.3), S3 Excellent (98.4), S4 Poor (48.4), S5 (Excel‑
lent (100.0), S6 Good (77.6), S7 Good (76.2), S8 Good (76.5), S9 Good (69.9), S10 Excellent (81.5), S11 Good (71.2), S12
Good (63.1), and S13 Good (63.5).) Our method effectively reduced the number of variables required for index calculation
compared with WQI methods of the MOE. Further, the reduced number of variables simplified the analysis process, reduced
analysis time, and enabled water quality assessment that reflected the water quality characteristics of the river to be evaluated.

Keywords Cluster analysis · Major variable · Principal component analysis · River tributaries · Spatial classification ·
Water quality management

Abbreviations Karakuş 2019). The increase in demand for water resources

WQI Water quality index and the imbalance in availability can pose a threat to human
WT Water temperature sustainable development. For this reason, each country is
evaluating the water footprint for water resource manage‑
ment; this is a concept that includes all water required from
Introduction product consumption to disposal, considering the entire
process of material production and service provision (Park
Rivers are a major source of drinking water and are also et al. 2020). In Korea, the life cycle assessment (LCA)
used for irrigation and industrial purposes; therefore, the method was used to calculate the water footprint and evalu‑ 1
prevention and management of river water pollution is cru‑ ation data of the rainwater exclusion system in considera‑
cial. Globally, climate change has led to a reduction in water tion of the environment. (Ahn et al. 2019). LCA is a tool to
resources, which is compounded by the increasing demand implement an environmental management system, and it is
for water because of rising urbanization (Haque et al. 2014; a technique to evaluate potential environmental impacts that
may occur in the manufacturing process of a product (Chen
et al. 2020). To achieve common performance for the circu‑
Editorial responsibility: Samareh Mirkia. lar economy from the management point of view of scarce
water resources, a strategy to increase economic advantage
* H. G. Kwon
hun7082@[Link] and reduce environmental water loss by performing LCA is
needed (Silvestri et al. 2021).
1
Nakdong River Environment Research Center, Reliable water quality data are required to prevent and
National Institute of Environmental Research, manage water pollution in rivers, and continuous water
Dalseong‑Gun, Daegu 43008, Republic of Korea

13
Vol.:(0123456789)
 2488 International Journal of Environmental Science and Technology (2023) 20:2487–2502

quality assessment must be conducted to monitor pollut‑ (Bharti and Katyal 2011), as the public can easily obtain
ants and water resource management based on spatial-scale and understand the relevant in-depth information (Naseem
water quality analyses (Wu et al. 2018). To assess water et al. 2021).
quality, several researchers have used methods, such as clas‑ When performing water quality evaluation of the sur‑
sification, correlation analysis, and time-series numerical vey site using WQI, the same water quality variables were
analysis of measured data (Singh et al. 2008; Sharma et al. applied to each stream in the previous studies (Shin et al.
2012). These methods have the advantage of not requiring 2018; Cho et al. 2021), whereas in this study, the water qual‑
expert knowledge about water quality or the environment. ity evaluation method that more quickly reflected the char‑
However, water quality assessment by time-series numerical acteristics of rivers using major water quality variables for
analysis on long-term measured data requires a cumbersome each stream was used. The applicability and feasibility of
and time-consuming data arrangement process. the results of this study were evaluated by selecting major
4
Principle component analysis (PCA), used in regional variables reflecting the quality of rivers using statistical tech‑
assessment to analyze major water quality variables, enables niques. We also compared the WQI calculated in this study
an in-depth analysis of river characteristics through complex with the WQI currently presented by the domestic Ministry
2 data matrix analysis (Tripathi et al. 2019). This method pro‑ of Environment. These results may serve as basis for simpli‑
vides a quick solution for the proper management of water fying the number of variables for water quality evaluation of
resources (Singh et al. 2004; Liu et al. 2011). Barakat et al. rivers in the future and for efficiently derive pollutants that
(2016) evaluated water quality variability of the Oum Er need to be preferentially managed. In addition, this study
Rbia and El Abid rivers in Morocco according to spatial may be used by policy makers for water quality improvement
(cluster) classification by using PCA and cluster analysis by region or by river.
(CA). These techniques are valuable for water resource man‑
agement as they can identify and help investigate the origin
of critical factors affecting rivers (Barakat et al. 2016; Varol Materials and methods
2020). Multivariate statistical analysis techniques, such as
PCA and CA, reflect temporal and spatial changes in water Study progress flow
quality and are used widely when evaluating water quality
for river management by assisting in the selection of com‑ Figure 1 shows the series of analysis procedures performed
ponents that affect river water quality (Gamble and Babbar- in this study. It was divided into three stages. First, clus‑
sebens 2012; Bora and Goswami 2016). ter analysis by survey sites, and major water quality vari‑
The combined impact of many different factors that ables were selected through multivariate statistical analysis.
characterize the water quality is complex; and so are the
3 challenges of classifying the significant parameters used to
measure the status of water resources quantitatively. There‑
fore, the water quality index (WQI) is considered a math‑
ematical tool that significantly simplifies these data sets and
provides a single classifying value that describes the water
quality status of water bodies or degree of pollution (Naseem
et al. 2021). WQI is a performance measurement that com‑
bines information from significant physical, chemical, and
biological parameters into a functional form, reducing large
amounts of data to a single number (Gradilla-Hernández
et al. 2019).
Using WQI, water quality is converted into a single score
for comprehensive assessment, helping the public and policy
makers to understand water quality (Terrado et al. 2010;
Upadhyay et al. 2011; Gupta et al. 2017). In WQI, a water
body is allocated a number between 0 and 100, which indi‑
cates its drinking water quality (Gradilla-Hernández et al.
2019). The WQI is the most effective way to express the
suitability of a water body, such as a river, as a water source
for human use (Ewaid et al. 2017; Khangembam et al. 2019).
Further, WQI is used as a policy-making tool for water qual‑
ity assessment by environmental monitoring organizations Fig. 1  Flowchart to illustrate how this study

13
International Journal of Environmental Science and Technology (2023) 20:2487–2502 2489

Second, the WQI (according to the difference in the appli‑ of the river could be determined (Pekey et al. 2004). CA
cation of water quality variables) was calculated and com‑ and PCA were standardized to z-scores to prevent errors
pared with RT-WQI. Third, water quality evaluation was between data owing to different units of measurement. An
performed using the WQI calculation method proposed in statistical software (IBM SPSS 24.0 Inc.) was used for
this study. data analysis.

Survey point and item

Cluster analysis
The Nam River is located in the southwest of South Korea.
It has a total flow length of 144.59 km and a basin area of 3 In this study, CA was performed for the spatial classi‑
467.5 ­km2. From the upper stream to the lower stream, the fication of nine water quality variables (water tempera‑
major tributaries of the Nam River include the Namgang, ture [WT], dissolved oxygen [DO], electrical conductiv‑
Ram, Obong, and Im streams, the Dongcheon River, Sinde‑ ity [EC], pH, biochemical oxygen demand [BOD], total
ung, Yang, and Migok streams, Youngcheon River, and the organic carbon [TOC], chemical oxygen demand [COD],
Euiryung and Haman streams. The survey points and vari‑ suspended Solid [SS], total nitrogen [TN], and total
ables of this study are shown in Table 1 and Fig. 2. The sur‑ phosphorus [TP]) of the 13 tributaries of the Nam River
vey points include the 13 water quality measurement points (Fig. 2). For that, we used monthly survey results from
at the main tributaries of the Nam River operated by the 2012 to 2020.
Ministry of Environment. Water quality data (2012–2020) CA collects and classifies variables with similar char‑
obtained from the Water Environment Information Sys‑ acteristics among multiple variables; it is a statistical
tem ([Link] were used. From the acquired analysis method that can identify the structure of data by
data, nine variables used in the RT-WQI were selected. In grouping them according to the homogeneous properties
2012, an artificial structure called a weir was installed in the of the variables. It classifies the data into clusters by using
Nakdong River, changing the water environment (Jo et al. the distance or the similarity between the variables of the
2022). Therefore, in this study, the monitoring results col‑ population, and it can be classified as a hierarchical or a
lected from 2012 to 2020, after the artificial structures were non-hierarchical method, according to the method used
installed, were applied to the water quality evaluation. to analyze the clusters. To classify the clusters as tempo‑
rally and spatially homogeneous based on the correlation
Statistical analysis of the data (Vega et al. 1998), hierarchical CA, in which
sequential clustering occurs, is used and is represented by
CA and PCA were applied for the spatial classification a dendrogram (McKenna 2003). The hierarchical cluster‑
of the measured data and selection of major water qual‑ ing method is used for classification; the Euclidean dis‑
ity variables. Therefore, various rivers can be classified tance (Otto et al. 1998) is used for measuring the distance
according to their characteristics, and the main water qual‑ between clustered variables; and Ward’s method (Fan et al.
ity variables representing the water quality characteristics

Table 1  Survey point status and survey items

Point # Measurement point name Area ­(km2) Survey count and items

S1 Namgang stream 160.0 Amount of data used to calculate the WQI: 108 per variable at each site
S2 Ram stream 264.0 Survey interval: one per month
S3 Obong stream 218.2 Survey items: nine items
WT (Water temperature), EC (Electrical conductivity), DO (Dissolved
S4 Im stream
oxygen), pH, BOD (Biochemical oxygen demand), COD (Chemical
S5 Dongcheon river 354.5 oxygen demand), TOC (Total organic carbon), TN (Total nitrogen),
S6 Sindeung stream 172.1 TP (Total phosphorus)
S7 Yang stream upstream 252.5
S8 Migok stream
S9 Yang stream downstream
S10 Youngcheon river upstream 205.5
S11 Youngcheon river downstream
S12 Euiryung stream 114.4
S13 Haman stream 155.2

13
2490 International Journal of Environmental Science and Technology (2023) 20:2487–2502

Fig. 2  Nam River watershed with 13 water quality measurement points. The first letter of all the words in the name of the measurement locations
has been capitalized in this study

2010), which can minimize information loss between clus‑ quality characteristics of the basin surveyed. Based on the
ters, is used for merging. CA results, PCA was performed on the water quality com‑
ponents of each cluster. The eigenvalue represents the mag‑
Principal component analysis nitude of the variance that can be explained by the principal
component, and as mentioned, an eigenvalue greater than
PCA is used to find the direction of scattered data, re-express 1.0 means that one principal component (PC) could explain
the data to facilitate understanding of the information in the more than one variable. We used the only PC with an eigen‑
data, and reduce the variability of variables. In other words, value of 1 in our analyses. With regard to the components
it is a statistical analysis method to find a new variable of the ten water quality variables, the eigenvalue obviously
(principal component), expressed as a linear combination represents the eigenvalue, and variability (%) is the ratio
of variables by using the correlation between multiple vari‑ that explains the variability of the PC with respect to the
ables (Sârbu and Pop 2005; Barakat et al. 2016). In PCA, the original variable. Cumulative (%) is the cumulative ratio
eigenvalue represents the magnitude of the variance that can used to explain the PC.
be explained by the principal component. If the eigenvalue
is greater than 1.0, it means that one principal component Water quality index
can explain more than one variable. Therefore, the principal
component is extracted based on principal components with WQI converts important water quality variables into a sin‑
an eigenvalue of 1.0 or greater (Varol et al. 2012; Park et al. gle index, which can be used as an easy-to-understand scale
2019). when evaluating the water quality of a target stream. Many
In this study, PCA was performed to select the main water researchers have used multivariate statistical techniques,
quality variables (WT, EC, DO, pH, BOD, COD, TOC, TN, such as PCA and CA, for selecting key variables for calcu‑
and TP) by CA that could represent the characteristics of lating WQI (Sutadian et al. 2016, 2018; Tripathi and Singal
each stream and PCA was used to extract the components 2019; Gradilla-Hernández et al. 2019). With regard to the
that could have the most significant influence on the water assessment criteria for the calculated WQI, we used five

13
International Journal of Environmental Science and Technology (2023) 20:2487–2502 2491

Table 2  WQI assessment criteria provided by the Korean Ministry of Environment

*WQI
Assessment details WQI Index grade

Clean water with almost no pollutants, suitable for recreational water activities at all times 80 to 100 Excellent
Relatively good water quality maintained, suitable for recreational water activities 60 to 79 Good
Generally, good water quality maintained, but pollutants are introduced from time to time that could affect 40 to 59 Poor
recreational water activities
Contaminated water owing to frequent inflow of pollutants, raising a caution for recreational water activities 20 to 39 Very Poor
Severely contaminated water, not suitable for recreational water activities 0 to 19 Unsuitable for
drinking or fish
culture

*WQI water quality index

grades suggested by the Ministry of Environment (Table 2). four in Cluster 1 (COD, TOC, TN, and TP), four in Cluster 2
The five grades were Excellent (80 to 100), Good (60 to (WT, DO, COD, and TOC), and five in Cluster 3 (WT, DO,
79), Poor (40 to 59), Very Poor (20 to 39), and Unsuitable EC, COD, and TOC).
for Drinking Use and Fish Culture (0 to 19). With regard to
the water quality variables for calculating WQI, the water
quality data (WT, EC, DO, pH, BOD, COD, TOC, TN, and Results and discussion
TP) measured at 13 measurement points from the Water
Environment Information System ([Link] Results of water quality analysis according to spatial
were used. classification
For quality assessment of the water of the Namgang
inflow tributaries, we calculated the annual WQI from 2012 CA was performed based on the area according to the land-
to 2020. The WQI was calculated as in Eq. 1 (Seo et al. use type in the river basin for the classification of the water
2021a, b) for seven factors, including WT, pH, DO, EC, quality concentration and spatial characteristics of the 13
TOC, TN, and TP, and by varying the main water quality survey points (Fig. 2). The CA results were classified into
variable for each group. The main water quality variables three clusters, and the survey points for each cluster were
applied in the calculation of WQI for each cluster were Cluster 1 (S1 to S5), Cluster 2 (S6 to S11), and Cluster 3
selected by PCA. (S12 and S13). The analyses result of the characteristics of
the basin for each survey point indicated that the clusters
(1)
√�� 2
F1 + F22 + F32 ∕3
� �
WQI = 100 − were formed by the influence of urbanization and human
activities (Fig. 4). In terms of the land-use ratio, the aver‑
where F1 is the fraction of the number of water quality vari‑
age urbanization/drying ratio was 2.2% for Cluster 1, 3.6%
ables that violate the criteria, F2 is the fraction of the total
for Cluster 2, and 5.7% for Cluster 3. The average ratio of
number of times the criterion is violated in all water qual‑
agricultural land was 11.7% for Cluster 1, 17.9% for Cluster
ity variable, and F3 is calculated by the sum of the values
2 17.9%, and 18.4% for Cluster 3, demonstrating a trend
obtained by fractionating the degree of violation of the crite‑
similar to the urbanization/drying ratio. As the ratio of the
ria for the water quality variable. Details are available from
forests in the basin (Cluster 1: 78.8%, Cluster 2: 68.3%, and
the Real-time Water Quality Information System ([Link]
Cluster 3: 63.8%) that represented the natural characteristics
www.k​ oreaw​ qi.g​ o.k​ r). The assessment for each survey point
of the basin showed an opposite trend from the previous
was conducted by varying the WQI calculation factors.
results, apparently urbanization and human activities were
The water quality index calculated using the RT-WQI
the crucial factors defining the characteristics of the water
provided by the Ministry of Environment was compared
basin. Seo et al. (2021a) identified water quality variables
with the index calculated based on the method proposed
that affect the spatial difference and concentration of water
in this study using the PCs for each Cluster (Table 3 and
quality by river through a study on the effect of spatial vari‑
Fig. 3). Seven water quality variables were applied to the
ability by river basin on the water quality of the Nakdong
RT-WQI analysis provided by the Ministry of Environment,
River in Korea.
including WT, pH, DO, EC, TOC, TN, and TP. However, the
The water quality characteristics were explained by divid‑
water quality variables used in our method of calculating the
ing the water quality variables into clusters and expressing
WQI (i.e., using the principal variables of each Cluster) were
the average and range of concentrations for each variable

13
 2492

13
Table 3  WQI scores of the stations in the Nam River watershed during the study period
Year Method S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13

2012 RT-WQI* 82.51 76.37 100.00 100.00 100.00 87.52 79.22 75.16 94.04 92.46 76.19 68.99 59.52
This study** 71.08 41.99 100.00 71.00 100.00 79.75 77.94 70.42 89.77 90.28 71.00 57.53 53.61
2013 RT-WQI 94.17 88.36 100.00 88.37 100.00 94.05 86.52 87.94 81.52 87.70 79.75 55.98 58.61
This study 56.63 70.89 100.00 42.11 100.00 80.04 78.96 79.98 69.92 90.28 70.30 64.82 61.47
2014 RT-WQI 93.94 82.56 100.00 88.37 100.00 93.80 83.80 87.41 74.52 75.85 69.40 56.82 57.88
This study 85.55 42.16 100.00 42.15 100.00 80.09 76.51 80.10 69.19 70.93 70.45 61.16 61.80
2015 RT-WQI 94.19 99.31 100.00 94.06 100.00 93.83 86.11 87.46 73.59 92.89 75.13 56.98 64.25
This study 100.00 71.10 100.00 41.76 100.00 79.90 77.62 80.10 68.36 90.16 69.72 60.74 69.21
2016 RT-WQI 94.19 94.17 94.19 100.00 100.00 88.26 81.31 81.81 74.23 87.72 81.75 64.53 66.26
This study 70.99 42.14 100.00 70.92 100.00 70.56 68.63 70.27 68.37 80.44 70.12 68.82 71.87
2017 RT-WQI 88.11 88.13 100.00 94.06 100.00 81.98 77.11 84.24 71.38 80.05 78.23 54.06 49.96
This study 41.84 41.56 100.00 41.92 100.00 70.23 75.86 79.68 58.93 70.83 70.20 57.62 59.75
2018 RT-WQI 94.06 94.06 94.19 94.03 100.00 87.42 83.18 78.15 73.53 80.98 84.51 57.41 60.14
This study 56.50 42.11 85.55 41.62 100.00 79.17 76.80 69.27 77.13 70.60 78.96 63.51 68.92
2019 RT-WQI 94.06 88.37 100.00 94.19 100.00 87.33 83.04 83.34 78.21 92.33 73.91 60.77 52.34
This study 42.12 42.16 100.00 41.93 100.00 78.81 76.55 78.90 67.55 89.99 69.71 66.31 61.21
2020 RT-WQI 88.23 94.19 100.00 94.19 100.00 93.52 79.90 86.05 75.49 82.02 81.82 67.85 54.58
This study 41.92 85.49 100.00 41.93 100.00 79.65 76.99 79.46 60.04 80.16 70.72 67.78 63.39
M ± SD RT-WQI 91.5 ± 4.2 89.5 ± 6.9 98.7 ± 2.6 94.1 ± 4.1 100 ± 0 89.7 ± 4.3 82.2 ± 3.1 83.5 ± 4.5 77.4 ± 6.9 85.8 ± 6.3 77.9 ± 4.7 60.4 ± 5.5 58.2 ± 5.3
This study 63 ± 20.7 53.3 ± 17.4 98.4 ± 4.8 48.4 ± 12.8 100 ± 0 77.6 ± 4.1 76.2 ± 3 76.5 ± 4.9 69.9 ± 9.2 81.5 ± 9 71.2 ± 2.9 63.1 ± 4.2 63.5 ± 5.7
t-test p-value 0.003 (○) 0.000 (○) 0.865 (●) 0.000 (○) – (●) 0.000 (○) 0.001 (○) 0.006 (●) 0.069 (●) 0.261 (●) 0.002 (○) 0.245 (●) 0.05 7(●)

The average water quality index (WQI) was calculated for nine years using two methods and conducting a t-test. ○: p < 0.05, ●: p > 0.05; M: mean; SD: standard deviation
*Variables used to calculate the RT-Water Quality Index (RT-WQI) are WT, pH, DO, EC, TOC, TN, and TP
**Variables used to calculate WQI in this study are (a) COD, TOC, TN, and TP (S1–S5), (b) WT, DO, COD, and TOC (S6–S11), and (c) WT, DO, EC, COD, and TOC (S12, S13
International Journal of Environmental Science and Technology (2023) 20:2487–2502
International Journal of Environmental Science and Technology (2023) 20:2487–2502 2493

Fig. 3  Comparison of the

assessment results for each
method used to calculate the
water quality index. Point S5 is
not shown because the assess‑
ment results are the same

13
2494 International Journal of Environmental Science and Technology (2023) 20:2487–2502

Fig. 4  Ratio of the land-use area in the basin to which the river expressed using the cluster analysis. The land-use types include
belongs. The upper right table shows the ratio of area according to urbanization/drying land, agricultural land, forest, grassland, wetland,
land use in the watershed at the survey sites, with the results being bare land, and fresh water

(Table 4 and Fig. 5). Interquartile range (IQR) was calcu‑ watershed to which a stream belongs can be distinguished by
lated by subtracting the third quartile by the first quartile, the influence of human activities, such as urbanization and
and the text (A, B, and C) on the box plot was used to char‑ agriculture, similar to the results of CA. Compared with the
acterize the homogeneity of the Scheffe test (analysis of other clusters, the water quality concentration of the basins
variance, ANOVA; p < 0.05). in Cluster 1 was superior because of its strong natural char‑
In the results shown in the box plots for each group, acteristics, whereas the water quality concentration of the
ANOVA results according to the Scheffe method also basins in Cluster 3 was inferior because of more urbanized
appeared to confirm the homogeneity between the groups. and agricultural lands. Other studies have suggested that
The WT was analyzed as 16.4 °C and 16.9 °C on average human activities affecting agricultural and urbanization of
for Clusters 1 and 3, and 14.4 °C on average for Cluster 2. land have a significant positive correlation with water quality
Among the investigated water quality variables, the mean variables of rivers (Giri and Qiu 2016; Lintern et al. 2018).
concentration, and the range of change in concentration of The results of analyzing clusters and water quality charac‑
electrical conductivity (EC), BOD, COD, TOC, TN and TP teristics appeared to suggest the need for the differentiation
were the highest in Cluster 3 with a wide range, followed by of the method and target item for water quality assessment
Clusters 2 and 1 (Fig. 6). With regard to DO, Clusters 1 and according to the land-use characteristics of the basin.
2 had similar mean values (11.0, 11.1 mg/L, respectively)
and ranges (6.7 to 17.0 mg/L, 5.7 to 17.5 mg/L, respec‑ Results of principal component analysis
tively), whereas Cluster 3 had a lower mean concentration
(10.1 mg/L). With regard to pH, similar concentrations and The results of the PCA indicated the cumulative explanatory
trends were observed for all clusters. With regard to the clas‑ power of Clusters 1 and 2 as 70% and 65%, respectively,
sification of the basin area by land use as shown in Fig. 4, with only three principal components, and the cumulative
the results show that the water quality characteristics of each explanatory power of Cluster 3 as 82%, which could be

13
 Table 4  Average concentration and concentration range of each water quality variable at survey points
Variables Cluster 1 Cluster 2 Cluster 3

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13

WT (°C) R 0–29.3 0.1–30.3 2.1–27.4 1.4–30.4 3.5–29.7 1.9–30.9 3.1–30.6 1.9–29.8 1.4–30.4 3.5–31 2.7–30.4 4.5–30 0.8–31.9
M ± SD 13 ± 8.5 13.6 ± 8.6 14 ± 7.7 14.6 ± 8.2 16.7 ± 7.5 15.6 ± 8.5 16.8 ± 8.6 15.7 ± 8.5 16.4 ± 8.8 17.3 ± 7.9 16.5 ± 8.7 17.1 ± 7.5 16.7 ± 8.9
EC (µs/cm) R 39–285 8–214 49–110 52–171 35–94 69–246 84–344 79–342 47–765 93–431 41–460 77–1852 103–637
M ± SD 108.3 ± 40.8 120.4 ± 35.6 75 ± 13.3 102.9 ± 28.2 59.9 ± 13.9 151.5 ± 36.4 220.9 ± 44.7 232.8 ± 48.7 210.6 ± 108 220.4 ± 67.9 224 ± 86.3 635.7 ± 382.8 319.6 ± 117.7
DO (mg/L) R 7.3–17 6.7–14.5 7.4–15.4 7.4–15.7 7.6–15.4 7.8–17.5 7–16.1 7.4–17.5 5.7–16.2 7.6–15.9 6.5–16.5 5.7–13.7 4.9–18.2
M ± SD 11.5 ± 2.5 10.8 ± 1.9 10.8 ± 2.2 11.2 ± 2.3 10.9 ± 2.2 11.7 ± 2.5 11.3 ± 2.4 11.6 ± 2.1 10 ± 2.8 11.3 ± 2 10.7 ± 2.6 8.9 ± 1.9 11.3 ± 3.8
pH R 6.6–8.7 6.2–9.2 7.1–8.1 7–8.6 7.1–8.5 7.1–8.9 7.1–9.1 7.1–8.7 6.7–9.1 7.2–8.9 0.8–9 6.4–9 7–9.4
M ± SD 7.9 ± 0.5 8.4 ± 0.3 7.5 ± 0.2 7.7 ± 0.3 7.8 ± 0.3 8 ± 0.4 8.1 ± 0.5 7.9 ± 0.4 7.9 ± 0.6 8.1 ± 0.4 7.9 ± 1.1 7.6 ± 0.4 8.1 ± 0.6
BOD (mg/L) R 0.1–2.7 0.1–4.3 0.3–0.8 0.4–1.8 0.3–1.2 0.4–2 0.5–2.7 0.4–2 0.3–3.2 0.3–1.7 0.4–4.8 0.3–3.9 0.7–8.8
M ± SD 0.9 ± 0.6 0.8 ± 0.7 0.5 ± 0.1 0.8 ± 0.3 0.7 ± 0.2 1 ± 0.4 1.3 ± 0.5 1 ± 0.4 1.4 ± 0.6 1 ± 0.3 1.6 ± 0.9 1.7 ± 0.8 3.5 ± 1.9
COD (mg/L) R 1.2–5.7 1.3–7.7 1–4.4 1.7–7.6 1.2–3.6 2.1–8.6 3.1–9.5 2.2–8.3 1.6–7.6 1.8–6.1 1.8–7.7 2.6–15.5 2.2–11.8
International Journal of Environmental Science and Technology (2023) 20:2487–2502

M ± SD 2.7 ± 1.1 3.4 ± 1.3 2.2 ± 0.7 3.4 ± 1 2.3 ± 0.6 4.4 ± 1.6 5.5 ± 1.4 4.4 ± 1.3 4.5 ± 1.5 3.4 ± 1 4 ± 1.3 7.1 ± 2.7 6.6 ± 2
TOC (mg/L) R 0.3–3.9 0.5–3.5 0.4–2.7 1–5.6 0.4–2.3 0.9–6 1.8–7.4 1.3–6 0.8–8.3 0.7–4.2 0.5–6.4 1.1–10.1 1.6–8.7
M ± SD 2.1 ± 0.8 1.8 ± 0.7 1.1 ± 0.4 2.1 ± 0.8 1.2 ± 0.4 2.9 ± 1.3 4 ± 1.2 3 ± 1.1 4.1 ± 1.5 2.1 ± 0.8 3.2 ± 1.3 5.5 ± 2.2 4.8 ± 1.5
TN (mg/L) R 0.921–2.675 0.87–2.679 0.377–1.266 0.693–2.752 0.706–2.186 0.56–1.996 0.397–3.284 0.256–2.851 0.329–4.277 1.117–3.263 0.309–2.914 0.737–7.811 0.797–5.243
M ± SD 1.802 ± 0.436 1.735 ± 0.457 0.756 ± 0.192 1.442 ± 0.39 1.14 ± 0.332 1.2 ± 0.31 1.187 ± 0.572 1.142 ± 0.488 1.133 ± 0.645 2.085 ± 0.4 1.35 ± 0.606 2.996 ± 1.323 2.637 ± 1.048
TP (mg/L) R 0.003–0.196 0.006–0.131 0.003–0.048 0.006–0.073 0.004–0.057 0.009–0.137 0.008–0.061 0.004–0.078 0.005–0.059 0.012–0.09 0.003–0.089 0.013–0.169 0.015–0.438
M ± SD 0.025 ± 0.03 0.036 ± 0.025 0.014 ± 0.009 0.026 ± 0.015 0.019 ± 0.013 0.037 ± 0.026 0.024 ± 0.014 0.027 ± 0.018 0.023 ± 0.011 0.043 ± 0.019 0.029 ± 0.017 0.047 ± 0.028 0.089 ± 0.069

The average concentration and the concentration range are tabulated for 108 water quality data measured monthly by survey point from 2012 to 2020
R: range; M: mean; SD: standard deviation. Range is indicated by using the standard deviation and the minimum and maximum concentrations of the water quality variables at each survey point,
with the average and standard deviation shown in the table

13
2495
2496 International Journal of Environmental Science and Technology (2023) 20:2487–2502

Fig. 5  Boxplot of the water quality measurement data from 2012 to 2020 for each group, confirming homogeneity according to spatial classification

13
International Journal of Environmental Science and Technology (2023) 20:2487–2502 2497

Fig. 6  Comparison of key water quality variables according to the that could explain all variables (WT, pH, EC, DO, BOD, COD, TOC,
cluster analysis. The larger the factor-loading value of the component, SS, TN, and TP)
the closer it is to the circle, thereby indicating a stronger influence

Table 5  Results of analyses to determine variability and eigenvalue of principal components (PC) by cluster

Division PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 PC9 PC10

Cluster 1 Eigenvalue 3.430 2.376 1.243 0.769 0.687 0.486 0.380 0.319 0.239 0.071
Variability (%) 34.297 23.763 12.433 7.686 6.874 4.863 3.796 3.186 2.394 0.709
Cumulative (%) 34.297 58.060 70.492 78.179 85.052 89.915 93.711 96.897 99.291 100.000
Cluster 2 Eigenvalue 3.573 1.634 1.333 0.967 0.794 0.595 0.498 0.313 0.159 0.134
Variability (%) 35.725 16.336 13.328 9.675 7.939 5.954 4.978 3.128 1.593 1.344
Cumulative (%) 35.725 52.061 65.389 75.064 83.003 88.957 93.936 97.064 98.656 100.000
Cluster 3 Eigenvalue 3.101 2.170 1.658 1.241 0.631 0.440 0.323 0.197 0.160 0.079
Variability (%) 31.012 21.696 16.579 12.412 6.310 4.396 3.235 1.967 1.600 0.793
Cumulative (%) 31.012 52.707 69.287 81.699 88.009 92.405 95.640 97.607 99.207 100.000

Bold values indicate eigenvalues > 1.0

Eigenvalue: eigenvalue of the principal component; Variability: total variance, which is the ratio that describes the variability of the original
variable; Cumulative: cumulative value of the total variance to describe the principal component

reduced to two principal components. As the number of vari‑ have the strongest influence in evaluating the water qual‑
ables decreases, the efficiency of PCA increases, indicating ity characteristics of the Cluster. In the case of PC2, TN
the value of the method proposed in this study, i.e., reducing (0.852) was selected. For PC1 of Cluster 2, WT (0.794),
a total of ten components to 3 and 4 PCs per Cluster (Table 5 DO (0.756), COD (0.854), and TOC (0.732) were selected,
and Fig. 6). and there was no variable corresponding to “strong” in
With regard to PCA, the factor-loading value for each PC2. For PC1 of Cluster 3, EC (0.889), COD (0.895), and
component was considered “strong” when it exceeded 0.7, TOC (0.882) were selected; for PC2, WT (0.790) and DO
“moderate” when it was between 0.7 and 0.5, and “weak” (0.849) were selected as principal variables. Cho et al.
when it was < 0.5. The results of PCA for each cluster and (2021) selected four out of ten major water quality vari‑
component are shown in Table 6. To select the principal ables using PCA to identify major variables contributing
component, the component corresponding to “strong” was to the water quality fluctuation characteristics of the South
selected among the factor-loading values of PC1 and PC2, Han River in Korea. Kim et al. (2007) reported that in
which had the highest explanatory power. determining the number of major water quality variables,
Among the PC1 variables of Cluster 1, COD (0.864), the determined variable was sufficient to explain the whole
TOC (0.794), and TP (0.718) were selected as the factor- when the cumulative % was 60% to 80% or more.
loading values corresponding to “strong”, that is, which

13
2498 International Journal of Environmental Science and Technology (2023) 20:2487–2502

Table 6  Principal component analysis by cluster

Cluster Variable Component
PC 1 PC 2 PC 3 PC 4

Cluster 1 WT 0.638 − 0.588 − 0.417

pH 0.045 0.446 − 0.521
DO − 0.603 0.640 0.382
EC 0.130 0.690 − 0.420
BOD 0.603 0.327 0.164
COD 0.864 0.167 − 0.019
TOC 0.794 0.255 − 0.132
SS 0.601 − 0.121 0.544
TN 0.148 0.852 0.063
TP 0.718 0.075 0.362
Eigenvalue 3.430 2.376 1.243
Variability (%) 34.297 23.763 12.433
Cumulative (%) 34.297 58.060 70.492
Cluster 2 WT 0.794 − 0.146 − 0.412
pH − 0.108 0.484 0.012
Fig. 7  Comparison of the water quality index assessment results and
DO − 0.756 0.299 0.406 the annual average water quality at point S1. The difference between
EC − 0.312 0.580 0.310 the existing method used for calculating the water quality index (RT-
BOD 0.479 0.517 0.238 WQI) and the calculation method proposed in this study was analyzed
COD 0.854 0.263 0.173
TOC 0.732 0.467 − 0.008
SS 0.527 − 0.165 0.533 Results of water quality index assessment
TN − 0.172 − 0.499 0.584
TP 0.679 − 0.350 0.437 After comparing the calculated WQIs (Table 3, Fig. 3), we
Eigenvalue 3.573 1.634 1.333 obtained similar results for most sites despite the differences
Variability (%) 35.725 16.336 13.328 in the assessment factors. However, the results for S1, S2,
Cumulative (%) 35.725 52.061 65.389 and S4 differed. The river water quality using the existing
Cluster 3 WT − 0.389 0.790 0.114 − 0.280 RT-WQI for S1 was evaluated as “Excellent” every year
pH − 0.379 − 0.303 0.661 − 0.332 from 2013 to 2016. However, the WQI calculation results
DO 0.031 − 0.849 0.398 − 0.040 applying the principal variables were evaluated differently
EC 0.889 − 0.002 − 0.204 − 0.106 by year (Very Poor, in 2013; Good, in 2014 and 2015; and
BOD 0.425 − 0.065 0.771 − 0.101 Poor, in 2016).
COD 0.895 0.233 0.117 − 0.221 The concentration of each water quality variable in the
TOC 0.882 0.197 0.023 − 0.299 same year as the WQI assessment is shown in Fig. 7. From
SS 0.028 0.387 0.39 0.590 2013 to 2016, which showed a large difference in the WQI
TN 0.473 − 0.488 − 0.172 0.561 assessment, the differences in concentration by year were
TP 0.175 0.496 0.466 0.477 identified clearly, and the same differences as in the previ‑
Eigenvalue 3.101 2.170 1.658 1.241 ous results were observed from 2018 to 2019. As the index
Variability (%) 31.012 21.696 16.579 12.412 is calculated based on factors applicable to all the basins
Cumulative (%) 31.012 52.707 69.287 81.699 in South Korea for RT-WQI, it will be difficult to reflect
the detailed characteristics of a specific basin. For S2 and
Bold and italic values represent strong and moderate loading, respec‑
tively S4, the WQI calculation results were analyzed differently
depending on whether principal variables were applied. As
a result of comparing the water quality concentration change
The selected principal components representing the by year and the calculation result (Fig. 7), it was determined
water quality variables at the survey points selected after that calculating the WQI by selecting the main variables
performing PCA included COD, TOC, TN, and TP for for each river would reflect the characteristics of the tar‑
Cluster 1; WT, DO, COD, and TOC for Cluster 2; and WT, get river in more detail. When the change in water qual‑
DO, EC, COD, and TOC for Cluster 3. ity by year and the WQI calculation result were compared,
the concentration of water quality variables showed (BOD

13
International Journal of Environmental Science and Technology (2023) 20:2487–2502 2499

0.5–1.5 mg/L, COD 2.4–3.0 mg/L, TN 1.647–2.162 mg/L, assessment for the water quality variables proposed in this
TP 0.020–0.065 mg/L, and TOC 1.3–2.3 mg/L) a change study, as shown in Fig. 8. For survey points classified as
according to the surveyed year. However, the RT-WQI cal‑ Cluster 1 (S1, S2, S3, S4, and S5), the average and annual
culation result for the same year was analyzed as Good, indi‑ assessment results at the S3 and S5 points were Excellent.
cating no change in water quality by year. Conversely, the The average assessment result was Good at S1, but the
WQI results of this study showed a similar trend from Excel‑ annual assessment result was Poor in 2019 and 2020 com‑
lent to Good according to the annual concentration change pared with that in 2016, indicating that the water quality
of water quality variables. Kang et al., (2019a) analyzed the was deteriorating. The assessment result at S4 was the worst
long-term measurement data of the Nam River in Korea. The (Very Poor) of all the survey points. At S2, upstream of S4,
authors suggested that monthly river water quality variability the assessment result was Poor on average, but the water
is large. In addition, a study that evaluated the water quality quality improved to Excellent in 2020, indicating that the
grade for Namgang reported that the yearly and seasonal water quality at S4 could have been affected partially by the
evaluation results showed similar trends to water quality upstream environment, but the extent of this influence was
variability (Kang et al. 2019b). According to Pejman et al. not absolute. Based on these results, the pollutants dispersed
(2009), when water quality is evaluated in a watershed, the across the basin at S4 must be monitored to enhance the
principal water quality variables reportedly appear different water quality. With regard to the survey points classified as
depending on the types of rivers distributed in the watershed. Cluster 2 (S6, S7, S8, S9, S10, and S11), the water quality
For some rivers (S1–S3), the assessment results were ana‑ was assessed as Good or Excellent, except for that of S9.
lyzed differently depending on the selection conditions of At S9, the water quality deteriorated from Good in 2018 to
the water quality variables; however, when compared with Poor in 2020. With regard to the survey points classified as
the concentration of the water quality variables by year, the Cluster 3 (S12 and S13), the water quality was assessed as
assessment results by the WQI applied in this study reflect Good on average; however, the WQI assessment indicated a
the water quality. value in the boundary between Poor and Good. This result
The results of the annual water quality assessment from points to the need for continuous water quality management
2016 to 2020 for each river or stream were incorporated in the future.
and compared with the nine-year average of water quality

Fig. 8  Water quality assessment results for each river basin and analyses of the change patterns in the river basin compared to the results of the
water quality assessment over the last five years. The assessment results over the last five years are grouped into previously analyzed clusters

13
 2500 International Journal of Environmental Science and Technology (2023) 20:2487–2502

Conclusion river will enable priority management of rivers that are more
vulnerable to pollution by establishing the direction of water
In this study, statistical techniques were used to select the quality and basin management for rivers.
main water quality variables of streams, and the WQI was
Acknowledgements This work was supported by a grant from the
calculated by applying the same water quality variables to National Institute of Environment Research (NIER), funded by the
streams with the same spatial classification. By comparing Ministry of Environment (MOE) of the Republic of Korea [grant num‑
the results obtained using the RT-WQI and those obtained ber NIER-2022-01-01-043].
by the WQI calculation method proposed in this study,
Author’s contribution All authors contributed to the planning and
the main water quality variables (COD, TOC, TN and TP structuring of the study. HGK wrote the manuscript and analyzed the
for Cluster 1; WT, DO, COD and TOC for Cluster 2; WT, data. CDJ conducted data collection. HGK reviewed the initial manu‑
DO, EC, COD, and TOC for Cluster 3) for each river were script and provided suggestions. All authors read and approved the
selected to assess water quality. The selected major vari‑ final manuscript.
ables can be used as representative items to evaluate the Funding This work was supported by a grant from the National Insti‑
characteristics of each river. As a result of the assessment, tute of Environment Research (NIER), funded by the Ministry of
WQI grade was assessment as S1 Good (63.0), S2 Poor Environment (MOE) of the Republic of Korea [grant number NIER-
(53.3), S3 Excellent (98.4), S4 Poor (48.4), S5 (Excellent 2022–01-01–043]. The funding agency had no role in study design; in
the collection, analysis and interpretation of data; in the writing of the
(100.0), S6 Good (77.6), S7 Good (76.2), S8 Good (76.5), report; and in the decision to submit the article for publication.
S9 Good (69.9), S10 Excellent (81.5), S11 Good (71.2),
S12 Good (63.1), and S13 Good (63.5). The water quality Data availability The datasets used and/or analyzed during the cur‑
of a river is highly dependent on the land-use status of rent study are available from the corresponding author on reasonable
request.
the basin, implying that the level of pollution would be
high in the basin subjected to extensive human activity. Code availability Not applicable.
Accordingly, such variables should preferably be consid‑
ered in the assessment of water quality reflecting the of Declarations
rivers and basins.
Conflict of interest The authors declare that they have no conflicts of
The limitation of the current study is that it only interest.
5
included the rivers with the water quality measurement
points operated by the Ministry of Environment. For more Ethical approval Not applicable.
accurate analyses than those of this study, a higher number
Consent to participate Not applicable.
of rivers should be included, and an additional review of
ionic substances would be required for the selection of Consent for publication Not applicable.
water quality parameters. And in the case of the current
analysis result, the method of this study can be applied Open Access This article is licensed under a Creative Commons Attri‑
even to rivers that have undergone the same water qual‑ bution 4.0 International License, which permits use, sharing, adapta‑
ity evaluation as the existing RT-WQI. However, efforts tion, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source,
to increase applicability through diverse case studies are provide a link to the Creative Commons licence, and indicate if changes
needed for rivers that show differences in some evaluation were made. The images or other third party material in this article are
results. included in the article's Creative Commons licence, unless indicated
The existing water quality improvement management otherwise in a credit line to the material. If material is not included in
the article's Creative Commons licence and your intended use is not
6 plan uses a single standard for all rivers and basins in South permitted by statutory regulation or exceeds the permitted use, you will
Korea, implying several limitations in terms of space, need to obtain permission directly from the copyright holder. To view a
time, and expense. River management policies should be copy of this licence, visit [Link]
established in a direction that can improve the water qual‑
ity of rivers distributed throughout the country (bottom-up
method) by establishing a management plan reflecting the References
characteristics of the target rivers by watershed and improv‑
ing the water quality. As can be seen from the results of this Ahn J, Park SW, Kim YI (2019) Development and application of inte‑
study, it is necessary to simplify the variables statistically grated evaluation index of rainwater drainage system based on the
life cycle assessment. J Korea Acad-Ind Coop Soc 20:1–7. [Link]
and scientifically for water quality evaluation and to estab‑ doi.​org/​10.​5762/​KAIS.​2019.​20.7.1
lish water quality evaluation methods according to land-use Barakat A, El Baghdadi M, Rais J, Aghezzaf B, Slassi M (2016)
and river characteristics. Therefore, in the future, calculat‑ Assessment of spatial and seasonal water quality variation of Oum
ing and assessing WQI based on the characteristics of each Er Rbia River (Morocco) using multivariate statistical techniques.

13
International Journal of Environmental Science and Technology (2023) 20:2487–2502 2501

Int Soil Water Conserv Res 4:284–292. [Link] Kim JH, Choi CM, Kim WI, Lee JS, Jung GB, Han KH, Ryu JS, Lee
iswcr.​2016.​11.​002 JT, Kwun SK (2007) Multi-variate statistical analysis for evalua‑
Bharti N, Katyal D (2011) Water quality indices used for surface water tion of water quality properties in Korean rural watershed. Korean
vulnerability assessment. Int J Environ Sci 2:154–173 J Environ Agric 26:17–24
Bora M, Goswami DC (2016) Water quality assessment in terms of Lintern A, Webb JA, Ryu D, Liu S, Waters D, Leahy P, Bende-Michl
water quality index (WQI): Case study of the Kolong River, U, Western AW (2018) What are the key catchment characteris‑
Assam, India. Appl Water Sci 7:3125–3135. [Link] tics affecting spatial differences in riverine water quality? Water
1007/​s13201-​016-​0451-y Resour Res 54:7252–7272. [Link]
Chen D, Kim J, Batagalle V, Kim B-S (2020) Impact evaluation of 72
water footprint on stages of drainage works. KSCE J Civ Environ Liu WC, Yu HL, Chung CE (2011) Assessment of water quality in a
Eng Res 40:225–231. [Link] subtropical alpine lake using multivariate statistical techniques
0225 and geostatistical mapping: a case study. Int J Environ Res Public
Cho YC, Choi HM, Ryu IG, Kim SH, Shin DS, Yu SJ (2021) Assess‑ Health 8:1126–1140. [Link]
ment of water quality in the lower reaches Namhan river by using McKenna JE (2003) An enhanced cluster analysis program with
statistical analysis and water quality index (WQI). J Korean Soc bootstrap significance testing for ecological community analy‑
Water Environ 37:114–127. [Link] sis. Environ Model Softw 18:205–220. [Link]
2021.​37.2.​114 S1364-​8152(02)​00094-4
Ewaid SH, Abed SA (2017) Water quality index for Al-Gharraf river, Naseem A, Ishak MIS, Ahmad MI, Umar K, Yusuff MSMd, Anees MT,
southern Iraq. Egypt J Aquat Res 43:117–122. [Link] Qadir A, Almanasir YKA (2021) Modification of the water quality
1016/j.​ejar.​2017.​03.​001 index (WQI) process for simple calculation using the multi-crite‑
Fan X, Cui B, Zhao H, Zhang Z, Zhang H (2010) Assessment of river ria decision-making (MCDM) method: a review. Water 13:905.
water quality in Pearl River Delta using multivariate statistical [Link]
techniques. Procedia Environ Sci 2:1220–1234. [Link] Otto M (1998) Multivariate methods. In: Kellner R, Mermet JM,
10.​1016/j.​proenv.​2010.​10.​133 Otto M, Widmer HM (eds.) Analytical Chemistry Wiley/VCH,
Gamble A, Babbar-Sebens M (2012) On the use of multivariate Weinheim, Germany, p 916
statistical methods for combining in-stream monitoring data Park DY, Kim SJ, Park HS, Chung SW (2019) Simulations of the
and spatial analysis to characterize water quality conditions in effect of control and phosphate loading on the reduction of algae
the White River basin, Indiana, USA. Environ Monit Assess biomass in Gangjeong-Goryong Weir. J Environ Impact Assess
184:845–875. [Link] 28:507–524
Giri S, Qiu Z (2016) Understanding the relationship of land uses and Park S, Lee M, Park K, An Y (2020) Calculating virtual water for
water quality in twenty first century: a review. J Environ Manag international water transactions: deriving water footprints in
173:41–48. [Link] South Korea. J Korea Water Resour Assoc 53:765–772. [Link]
Gradilla-Hernández MS, de Anda J, García-González A, Meza-Rod‑ doi.​org/​10.​3741/​JKWRA.​2020.​53.S-​1.​765
ríguez D, Montes CY, Perfecto-Avalos Y (2019) Multivariate Pejman AH, Bidhendi GRN, Karbassi AR, Mehrdadi N, Bidhendi
water quality analysis of Lake Cajititlán. Mexico Environ Monit ME (2009) Evaluation of spatial and seasonal variations in
Assess 192:5. [Link] surface water quality using multivariate statistical techniques.
Gupta N, Pandey P, Hussain J (2017) Effect of physicochemical and Int J Environ Sci Technol 6:467–476. [Link]
biological parameters on the quality of river water of Narmada, BF033​26086
Madhya Pradesh, India. Water Sci 31:11–23. [Link] Pekey H, Karakaş D, Bakoğlu M (2004) Source apportionment of
1016/j.​wsj.​2017.​03.​002 trace metals in surface waters of a polluted stream using multi‑
Haque MM, Rahman A, Hagare D, Kibria G (2014) Probabilis‑ variate statistical analyses. Mar Pollut Bull 49:809–818. [Link]
tic water demand forecasting using projected climatic data doi.​org/​10.​1016/j.​marpo​lbul.​2004.​06.​029
for Blue Mountains water supply system in Australia. Water Sârbu C, Pop HF (2005) Principal component analysis versus fuzzy
Resour Manag 28:1959–1971. https:// ​ d oi. ​ o rg/ ​ 1 0. ​ 1 007/​ principal component analysis a case study: the quality of Dan‑
s11269-​014-​0587-1 ube water (1985–1996). Talanta 65:1215–1220. [Link]
Jo CD, Lee CG, Kwon HG (2022) Effects of multifunctional weir con‑ 10.​1016/j.​talan​ta.​2004.​08.​047
struction on key water quality indicators: a case study in Nakdong Seo MJ, Heo JH, Kim YS (2021a) Present and potential future criti‑
River, Korea. Int J Environ Sci Technol. [Link] cal source areas of nonpoint source pollution: a case of the Nak‑
s13762-​022-​03973-8 dong river watershed, South Korea. J Environ Sci Pollut Res Int
Karakuş CB (2019) Evaluation of water quality of Kızılırmak River 28:45676–45692. [Link]
(Sivas/Turkey) using geo-statistical and multivariable statistical Seo JY, Won JG, Choi JH, Kim SD (2021) Major watershed charac‑
approaches. Environ Dev Sustain 1–35 teristics influencing spatial variability of stream TP concentra‑
Kang DH, So YH, Park KD, Kim IK, Kim BW (2019a) Monthly vari‑ tion in the Nakdong river basin. J Korean Soc Water Environ
ation analysis of BOD, COD and TOC using long-term observa‑ 37:204–216. [Link]
tion data in the downstream of the Nam river. J Environ Sci Int Sharma A, Singh AK, Kumar K (2012) Environmental geochemistry
28:571–579. [Link] and quality assessment of surface and subsurface water of Mahi
Kang DH, So YH, Park KD, Kim IK, Kim BW (2019b) Water quality river basin, western India. Environ Earth Sci 65:1231–1250
level of the living environmental standards and regression analysis Shin SH, Im TH, Kim SH, Shin DS, Kwon HG (2018) Considera‑
in the major tributaries of the Nakdong river. J Environ Sci Int tion in applying korean-comprehensive water quality index(K-
28:357–369. [Link] CWQI) to assessment of water quality in the major tributaries
Khangembam S, Kshetrimayum KS (2019) Evaluation of hydrogeo‑ of Gumho-river basin. J Korean Soc Hazard Mitig 18:455–465.
chemical controlling factors and water quality index of water [Link]
resources of the Barak valley of Assam, Northeast India. Ground‑ Silvestri L, Forcina A, Di Bona G, Silvestri C (2021) Circular economy
water Sustain Dev 8:541–553. [Link] strategy of reusing olive mill wastewater in the ceramic industry:
02.​001 how the plant location can benefit environmental and economic

13
2502 International Journal of Environmental Science and Technology (2023) 20:2487–2502

performance. J Cleaner Prod 326:129388. [Link] index: a case study of river Ganga India. Ecol 96:430–436.
1016/j.​jclep​ro.​2021.​129388 [Link]
Singh AK, Mondal GC, Kumar S, Singh TB, Tewary BK, Sinha A Upadhyay R, Dasgupta N, Hasan A, Upadhyay SK (2011) Managing
(2008) Major ion chemistry, weathering processes and water water quality of River Yamuna in NCR Delhi. Phys Chem Earth
quality assessment in upper catchment of Damodar River basin, 36:372–378. [Link]
India. Environ Geol 54:745–758. [Link] oi.​o rg/​1 0.​1 007/​ Varol M (2020) Use of water quality index and multivariate statistical
s00254-​007-​0860-1 methods for the evaluation of water quality of a stream affected
Singh KP, Malik A, Mohan D, Sinha S (2004) Multivariate statistical by multiple stressors: a case study. Environ Pollut 266:115417.
techniques for the evaluation of spatial and temporal variations [Link]
in water quality of Gomti river (India)—A case study. Water Res Varol M, Gökot B, Bekleyen A, Şen B (2012) Spatial and temporal
38:3980–3992. [Link] variations in surface water quality of the dam reservoirs in the
Sutadian AD, Muttil N, Yilmaz AG, Perera BJC (2016) Develop‑ Tigris river basin, Turkey. CATENA 92:11–21. [Link]
ment of river water quality indices—A review. Environ Monit 10.​1016/j.​catena.​2011.​11.​013
Assess 188:58 Vega M, Pardo R, Barrado E, Debán L (1998) Assessment of seasonal
Sutadian AD, Muttil N, Yilmaz AG, Perera BJC (2018) Develop‑ and polluting effects on the quality of river water by exploratory
ment of a water quality index for rivers in west Java Province, data analysis. Water Res 32:3581–3592. [Link]
Indonesia. Ecol 85:966–982. [Link] S0043-​1354(98)​00138-9
2017.​11.​049 Wu Z, Wang X, Chen Y, Cai Y, Deng J (2018) Assessing river water
Terrado M, Barcel D, Tauler R, Borrell E, de Campos S (2010) quality using water quality index in Lake Taihu Basin, China. Sci
Surface water quality indices for the analysis of data generated Total Environ 612:914–922. [Link]
by automated sampling networks. Trends Anal Chem 29:40–52 2017.​08.​293
Tripathi M, Singal SK (2019) Use of principal component analysis
for parameter selection for development of a novel water quality

13