Environ Monit Assess (2019) 191: 505

https://doi.org/10.1007/s10661-019-7652-4

Performance of Seven Water Quality Indices (WQIs)

in a Mediterranean River
Ioanna Zotou & Vassilios A. Tsihrintzis &
Georgios D. Gikas

Received: 11 December 2018 / Accepted: 15 July 2019 / Published online: 24 July 2019
# Springer Nature Switzerland AG 2019

Abstract The comparative performance of seven wide- and the subsequent qualitative classification of the water
ly used Water Quality Indices (WQIs) is evaluated in a body, the most important quality parameters per meth-
fluvial water body. The computation of the individual odology were determined through the calculation of the
WQIs was based on bi-weekly measurements of eleven correlation coefficient between each water quality pa-
water quality parameters in Vosvozis River situated in rameter and the final WQI value. The results showed
Rhodope Prefecture, North Greece. The sampling cam- that Bhargava’s Index along with NSF WQI tend to
paign lasted 16 months (from August 2005 to Novem- classify water bodies into superior quality classes,
ber 2006), and over this period, the individual parame- Prati’s Index into slightly lower categories, whereas
ters were measured at three gauging stations along the Oregon and CCME WQIs were found to be relatively
river. The indices which were applied were the Prati’s “stricter,” giving results ranging between the lowest
Index of Pollution, Bhargava’s Index, Oregon WQI, classes of the qualitative ranking. These findings con-
National Sanitation Foundation (NSF) WQI, Canadian firm those emerged by our previous studies while they
Council of Ministers of Environment (CCME) WQI, are also partially in line with similar studies comparing a
Dinius’ Second Index, and the Weighted Arithmetic smaller number of indices. CCME WQI is suggested as
WQI. Apart from the computation of the above WQIs the most appropriate, among the examined indices,
since it is both “conservative” and adequately “sensi-
tive” to reflect changes in water quality.
I. Zotou : V. A. Tsihrintzis (*)
Centre for the Assessment of Natural Hazards and Proactive
Keywords Water quality index . Physicochemical
Planning & Laboratory of Reclamation Works and Water
Resources Management, School of Rural and Surveying parameters . Water quality assessment . Surface water
Engineering, National Technical University οf Athens, 9 Iroon bodies . Qualitative classification . Water pollution .
Polytechniou St., Zografou, 157 80 Athens, Greece Vosvozis River
e-mail: [email protected]
e-mail: [email protected]
e-mail: [email protected]
Introduction
I. Zotou
e-mail: [email protected]
Surface water bodies, which are the primary source of
G. D. Gikas fresh water for most human activities, have been sub-
Laboratory of Ecological Engineering and Technology, jected to severe environmental pressures over the last
Department of Environmental Engineering, School of
Engineering, Democritus University of Thrace, 67100 Xanthi,
decades, as a result of both natural and anthropogenic
Greece causes. Factors such as urbanization, population growth,
e-mail: [email protected] climatic changes, and intensification of both agriculture
505 Page 2 of 14 Environ Monit Assess (2019) 191: 505

and industry tend to lead gradually to the degradation of quality data required in the process, which in another
the existing surface water systems, both qualitatively case would need to be evaluated separately, is incorpo-
and quantitatively (Yogendra and Puttaiah 2008; rated into a single and easily manageable result of the
Akoteyon et al. 2011; Boskidis et al. 2010). Because total water quality. Besides, combining or comparing the
of the above, the systematic monitoring and evaluation quality data (measurements of the water quality vari-
of the qualitative and quantitative status of water bodies ables) is becoming feasible through the normalization
has become a crucial issue in recent years. Although process that a WQI methodology provides. Finally, the
quantitative assessment of water may be implemented in dependence of the qualitative classification from the
a relatively easy way simply by determining the total water use is an issue which is also addressed thanks to
water volume stored in a reservoir or the water discharge the development of “specific consumption” WQIs
at a certain position of a river, the determination of the intending to evaluate water quality for specific water
qualitative status of water is more complex. This is uses.
mainly due to the fact that water quality characterization Moreover, the concrete computational framework
is strongly related to the water use, i.e., a certain water that a WQI provides may enable comparisons between
sample may appear to be of “acceptable quality” for the qualitative conditions in different water systems
irrigation or industrial use but not for drinking purposes. (e.g., lakes, rivers, groundwaters, and coastal areas) or
In addition, quality is a function of all the different in different positions of the same water body. Addition-
substances which are found in the water at different ally, it allows the examination of the quality status of a
concentrations (Abbasi and Abbasi 2012), and whose given water body at different times over a certain period,
importance in the water quality evaluation process dif- facilitating, thus, the detection of eventual degradation
fers depending on the water use. or improvement trends (Abbasi and Abbasi 2012;
Thus, it becomes clear that a detailed record of all the Almeida et al. 2008). Independently of which method-
individual water quality parameters and substance con- ology is selected to be used, the implementation of a
centrations could not lead to an easily manageable and WQI includes the following steps: (a) selection of the
comprehensible result of the overall quality status of a quality parameters which will be taken into consider-
given water sample, while, at the same time, it could be ation; (b) normalization of these parameters, through
understood only from experts (Abbasi and Abbasi 2012; appropriate mathematical functions defined by each
Alexakis et al. 2016). Furthermore, issues arise during methodology, in order to make them comparable; (c)
the qualitative evaluation, such as the difficulty in com- assignment of a weighting factor to each parameter
bining quality parameters of different importance, units based on its relative importance to the overall water
of measurements, and concentration levels in order to quality; (d) aggregation of the normalized concentration
derive a single result of the quality status, which make values of the individual parameters to produce a single
the process even more complex. result of the overall quality which is usually expressed
In this context, Water Quality Indices (WQIs) have on a scale from 1 to 100 (Zotou et al. 2018).
received much attention in the last decades, since they Beginning with Horton’s Index (Horton 1965), sev-
can provide a simplified view of the quality status of a eral WQIs have been developed over the last decades.
water body, by attributing to it a single numerical value Differences that appear among them are related either to
(score) which reflects the total water quality. Based on their nature, e.g., whether they are physicochemical or
appropriate classification schemes, determined by each biological indices, or to subjects concerning the process
individual methodology, these scores are categorized in of their development, e.g., the number of quality param-
specific quality classes (often from 1 to 5) each express- eters taken into account, the weightage assigned to each
ing a certain quality status, such as “very bad,” “bad,” parameter according to its relative importance, the way
“medium,” “good,” or “excellent” water quality. WQIs of calculating the values of individual subindices and
present the potential to address the complexity charac- the value of the overall water quality index. Moreover,
terizing the nature of water quality evaluation process, different indices have been developed based on the
either in the computational part of the process or in the purpose of their application and the type of water use
part related to the facilitation of public awareness they are to evaluate. The common target of all the WQIs
through simple and comprehensible results. Through published following Horton’s index has been to reduce
the utilization of WQIs, the large number of water the subjectivity factor which enters into the individual
Environ Monit Assess (2019) 191: 505 Page 3 of 14 505

steps of WQI development process and which consti- until now, no WQIs have been developed solely for
tutes the main drawback of the latter. application in the Mediterranean rivers. This region
Many researchers, on a global scale, have elaborated presents particular characteristics regarding its aquatic
WQI studies in order to investigate the usefulness of systems compared to the rest of Europe (e.g., Northern
these tools and the reliance they provide under different Europe). Factors such as the intensely uneven distribu-
local conditions (e.g., Yisa and Jimoh 2010; Alobaidy tion of precipitation throughout the year, the higher
et al. 2010; Chauhan and Singh 2010; Kumari and Rani temperatures and evapotranspiration, the different land-
2014; Al-Mutairi et al. 2014; Shah and Joshi 2017; scape, and the characteristics of the Mediterranean flu-
Haque et al. 2015; Moyel and Hussain 2015; Deshmukh vial water bodies themselves (e.g., relatively small size
and Aher 2016; Ewaid 2016; Mitra and Reddy 2016; and intermittent flow) are among the conditions which
Trikoilidou et al. 2017; Gikas 2017; Raju and Singh generate specific “natural conditions” in Mediterranean
2017; Yaseen et al. 2018; Xia et al. 2018; Adimalla water systems. This, in turn, may dictate a different
et al. 2018). Fewer scientists have sought to implement approach concerning the application of WQIs and the
different WQI methodologies in a certain water body in water quality evaluation, in general. Thus, the investi-
order to evaluate their comparative performance (e.g., gation of the response of these indices in the Mediterra-
Pesce and Wunderlin 2000; Debels et al. 2005; Hashim nean aquatic systems, as a first step, and then their
et al. 2015; Darvishi et al. 2016; Alexakis et al. 2016; amendment, if needed, to suit specific conditions are
Zotou et al. 2018). of significant importance.
The present study attempts to examine the compara-
tive efficiency of different WQI methodologies by ap-
plying seven widely used indices in a fluvial water body Materials and methods
in Rhodope Prefecture, North Greece. In particular, the
implementation of the above methodologies was based Area of application
on bi-weekly water quality data, collected by Boskidis
et al. (2010) over a 16-month monitoring period (August Vosvozis is an ephemeral river, located in Rhodope
2005 to November 2006), measured at three gauging Prefecture, Thrace District, North Greece. Administra-
stations in Vosvozis River. The following WQIs were tively, it falls into the Region of Eastern Macedonia and
applied and evaluated in this work: Prati’s Index of Thrace and, more specifically, into Rhodope Regional
Pollution (Prati et al. 1971), Bhargava’s Index Unit. Vosvozis springs from Rhodope Mountains and
(Bhargava 1983), Oregon WQI (Cude 2001), National drains into Ismarida lagoon, a wetland of International
Sanitation Foundation (NSF) WQI (Brown et al. 1970), Importance, which is protected by the Ramsar Conven-
Canadian Council of Ministers of Environment tion. The total length of Vosvozis River is 40 km, while
(CCME) WQI (CCME 2001), Dinius’ Second Index its total drainage area is about 365 km2 (Boskidis et al.
(Dinius 1987), and the Weighted Arithmetic WQI 2010). Within the broader area of Vosvozis watershed,
(Tyagi et al. 2013). Apart from the application of the the terrain varies, and includes a mountainous area in the
individual WQIs in the water body being examined in North, which covers approximately 49% of the total
this paper, the “most important” quality parameters per surface, while the south part is a plain agricultural area
methodology were determined through the calculation where several crops such as wheat, cotton, and tobacco
of the correlation coefficient between each individual are cultivated (Boskidis et al. 2010).
parameter and the final index value of each methodolo- The climate varies depending on elevation. In partic-
gy applied, in an effort to find the most sensitive param- ular, the coldest month is January, with a mean monthly
eter of each WQI. Thus, the main objectives of this work temperature of 2.1 °C and 4.5 °C for the mountainous
were (a) to comparatively evaluate the performance of and plain areas of the watershed, respectively, while the
the seven WQIs in a Mediterranean fluvial water body; warmest month is August with mean temperatures 23.2
(b) to demonstrate the “most important parameters” per °C and 24.3 °C, for the north and south parts of the study
methodology; and (c) to evaluate the quality status of the area, respectively. Similar differences are observed in
given body and detect eventual trends or important the annual precipitation depth. In the north-mountainous
discrepancies between the water quality conditions of part of the watershed, the mean annual rainfall is ap-
the different monitoring stations. It should be noted that proximately 812.7 mm and in the south 554.1 mm
505 Page 4 of 14 Environ Monit Assess (2019) 191: 505

(Boskidis et al. 2010). The dry period in Vosvozis lagoon. The exact locations of the three gauging stations
watershed lasts 5 months (from June to October), where- within the watershed are illustrated in Fig. 1.
as for the rest of the year, the area is characterized by wet The physicochemical parameters used to compute
conditions due to the comparatively higher rainfall the individual WQIs were temperature (T), electrical
depth. For the period being examined in the present conductivity (EC), dissolved oxygen (DO), biochemical
study (August 2005 to November 2006), the rainfall oxygen demand (BOD5), chemical oxygen demand
depth was recorded at 982.1 mm and 582.0 mm in the (COD), pH, nitrite nitrogen (NO2−-N), nitrate nitrogen
mountainous and plain areas, respectively, and is char- (NO3−-N), ammonia nitrogen (NH4+-N), total phospho-
acterized as a “wet period.” rus (TP), and total Kjeldahl nitrogen (TKN). It is noted,
The study area has been subjected to pressures due to though, that each individual WQI incorporates a differ-
both point and non-point sources of pollution associated ent number and combination of the above quality pa-
with human activities taking place in the watershed. The rameters. The quality parameters included in each index
main point source of pollution is the wastewater treat- are summarized in Table 1.
ment plant of the city of Komotini, which releases For a detailed description of the monitoring survey,
effluents into Vosvozis River. On a daily basis, 10,500 one can refer to Boskidis et al. (2010). Briefly, the
m3 of effluents discharge into the stream, with mean values and concentrations of T, EC, DO, and pH were
concentrations in nutrients of 0.2 mg/L for total phos- determined through in situ measurements at the three
phorus (TP) and 1.3 mg/L for nitrate nitrogen (NO3−-N). gauging stations using WTW series instruments. In
As for the non-point sources of pollution in the area, addition, water samples were collected at the same
agricultural activity constitutes the most important cause positions, and were immediately transported to the lab-
of water quality degradation in Vosvozis River, due to oratory and analyzed for BOD5, COD, NO2−-N, NO3−-
the transport of nutrients into the stream through runoff N, NH4+-N, TP, and TKN, based on standard methods
(Boskidis et al. 2010). Vosvozis flows into Ismarida (Boskidis et al. 2010).
lagoon, which has also been subjected to severe pres-
sures as a result of agricultural runoff, since significant
quantities of fertilizers and other constituents end up in Application of the methodology
it, either directly or through the flow of Vosvozis
(Pisinaras et al. 2007). As mentioned above, the seven WQIs which were ap-
plied in this study in order to evaluate the water quality
Water quality monitoring conditions in Vosvozis River were the Prati’s Index of
Pollution, Bhargava’s Index, Oregon WQI, NSF WQI,
In order to apply the individual methodologies being CCME WQI, Dinius’ second Index, and the Weighted
examined in the present work, eleven water quality Arithmetic WQI. The selection of the above methodol-
parameters were taken into consideration, which were ogies was based on the compatibility of the type of water
measured during a previous research conducted by use they evaluate and the availability of the appropriate
Boskidis et al. (2010), on a bi-weekly basis over a 16- data for their implementation. Among them, only the
month period (from August 2005 to November 2006). Weighted Arithmetic WQI serves a different purpose,
Briefly, for the implementation of the monitoring sur- since it evaluates the suitability of water exclusively for
vey, three gauging stations (GS) were installed at differ- drinking purposes. Therefore, this index is not taken into
ent positions along the river, each one reflecting the comparison against the rest of the methodologies but
characteristics of each part of the drainage area as well instead its application in the present study has mostly a
as the related impact on the qualitative status of the research character.
corresponding reach. Therefore, station GS1 reflects In the following sections, the main characteris-
the quality conditions in the river in absence of any type tics of the applied WQIs, as well as some of the
of human activity, station GS2 intends to examine the main assumptions made during their implementa-
impact of the agricultural activity as well as of the tion, are briefly described. For further details on
wastewater treatment plant operating nearby, whereas the computational framework of the individual
station GS3, situated upstream of Ismarida lagoon, aims methodologies, one can refer to the respective
at indicating the quality of water finally flowing into the publications.
Environ Monit Assess (2019) 191: 505 Page 5 of 14 505

Fig. 1 Locations of the three gauging stations GS1, GS2, and GS3 in the Vosvozis watershed

Prati’s Index of pollution each indicating “excellent,” “acceptable,” “slightly pol-

luted,” “polluted,” and “heavily polluted” conditions,
This index was established by Prati et al. (1971) and respectively. It is noted that in the present study, for
aims at the evaluation of water quality in surface water reasons of compatibility among the various methodolo-
bodies for general water use. Contrary to the majority of gies, the abovementioned classification scale was
the indices examined in this study, Prati’s Index is a inversed so that the 5th class corresponds to the “excel-
“pollution index” or an “ascending scale index,” which lent” quality condition and the 1st class to the “heavily
means that the numerical index value increases with the polluted” one. The sub-index functions for each param-
degree of pollution. Prati’s Index takes into consider- eter along with the classification scheme of the method-
ation thirteen water quality parameters, whose concen- ology are described in detail by Prati et al. (1971).
trations, through appropriate mathematical functions
(sub-index functions), are transformed into levels of Bhargava’s WQI
pollution. The overall index value is determined by
computing the arithmetic mean of the individual sub- Bhargava’s (1983) WQI, which was created by
index values. The index ranges from 0 to 14, with higher Bhargava in 1983, evaluates the overall water quality
values expressing higher pollution levels. Based on the in a surface water body, including its suitability for (a)
overall index value, the water body being examined is swimming; (b) public water supplies; (c) agriculture; (d)
categorized into a certain pollution class from 1 to 5, industry; and (e) fish culture, wildlife, boating, and other
505 Page 6 of 14 Environ Monit Assess (2019) 191: 505

Table 1 Water quality parameters included in each index and recommended concentration values used in the computation of CCME and
Weighted Arithmetic WQIs

Parameter Prati’s Bhargava’s Oregon NSF CCME WQI Dinius’ Weighted Arithmetic WQI
Index WQI WQI WQI (recommended second (recommended concentration)
concentration) Index

T – ✓ ✓ – – –
DO ✓ ✓ ✓ ✓ ✓ 6 mg/L ✓ ✓ 5 mg/L
pH ✓ – ✓ ✓ ✓ 6.5–9.0 mg/L ✓ ✓ 6.5–9.5
EC – ✓ – – – ✓ ✓ 2500 μS/cm
BOD5 ✓ ✓ ✓ ✓ ✓ 4 mg/L ✓ ✓ 3 mg/L
COD ✓ – – – – – ✓ 30 mg/L
NO2−-N – – – – – – ✓ 3 mg/L
NO3−-N ✓ – – ✓ – ✓ ✓ 50 mg/L
NH4+-N ✓ ✓ – – ✓ 0.6 mg/L – ✓ 0.5 mg/L
(NO2−-N) + (NO3−-N) – – – – ✓ 10 mg/L – –
(NO3−-N) + (NH4+-N) – – ✓ – – – –
TKN – – – – – – ✓ 1 mg/L
TP –` – ✓ ✓ ✓ 0.015 mg/L – ✓ 0.05 mg/L

non-contact recreation. This index integrates a sum of five categories, i.e., very poor, poor, fair, good, and
nine quality parameters in order to classify a surface excellent (Cude 2001).
water body. The main characteristic of Bhargava’s meth-
odology is that it incorporates a different combination of
parameters in the computation of the WQI for each NSF WQI
individual type of water use, indicating that not all
quality parameters are equally important for all uses. NSF WQI was created by Brown et al. (1970) and
Then, the overall WQI is calculated as the arithmetic aims at the evaluation of water quality for general
mean of the indices representing each water use. The use. Its application is based on the assignment of a
final index value ranges from 0 to 100 and, based on the weighting factor to each quality parameter included
classification system proposed by the author, it is attrib- in the index, according to its relative importance for
uted to a quality class from 1 to 5, each expressing the overall quality. The concentrations of these pa-
“unacceptable,” “poor,” “satisfactory,” “good,” and “ex- rameters, after being normalized through sub-index
cellent” water quality, respectively (Bhargava 1983). functions defined by the methodology, are aggregat-
ed to produce the final index as a weighted arith-
Oregon WQI metic mean. The final NSF value ranges from 0 to
100, with 100 representing the best water quality,
Oregon WQI evaluates the quality of water for general and categorizes the water body into a certain quality
recreational use, including fishing and swimming, by class from 1 to 5 based on the classification scheme
integrating measurements of eight water quality param- proposed by the methodology (Brown et al. 1970;
eters. The concentrations of these parameters, being Abbasi and Abbasi 2012). According to this classi-
expressed in different units of measurement, are trans- fication, the water quality condition is categorized as
formed into unitless sub-index values through appropri- “very bad,” “bad,” “medium,” “good,” or “excel-
ate functions (sub-index functions). Sub-index values lent”. In this study, the calculation of the index
are then combined through an aggregation function to was implemented by using an appropriate calculator
produce the final index, ranging from 10 (worst quality) freely available at http:/www.water-research.net/index.
to 100 (best quality). According to this score, Oregon php/water-treatment/water-monitoring/monitoring-the-
WQI classifies the quality status of the water body into quality-of-surfacewaters (accessed on May 31, 2019).
Environ Monit Assess (2019) 191: 505 Page 7 of 14 505

CCME WQI between 0 and 100. However, this value does not
correspond to a certain quality class but to a differ-
CCME WQI assesses the condition of aquatic life in ent one for each water use being considered. There-
a water body. For this purpose, it incorporates three fore, for simplification, the classification of
measures, i.e., scope, frequency, and amplitude, to Vosvozis River was made taking into account only
compute the final CCME index, which ranges from one of the six water uses, namely, that of fish life.
0 (worst quality scenario) to 100 (best quality sce- This assumption was based on the determinant role
nario). The above measures represent (a) the number of the above use compared to the rest of the water
of parameters whose water quality standards are not uses for which the water body is exploited. At the
met (scope); (b) the frequency with which these same time, due to the absence of a uniform classifi-
objectives are not met (frequency); and (c) the cation system, a different number of quality classes
amount by which the objectives are not met (ampli- is shaped for each type of water use, which makes
tude). They are determined taking into account ob- difficult the comparison with the rest of the meth-
jective values defined by the user. Depending on the odologies. Therefore, in the case of fish life water
final index score, the quality condition of the water use, where six quality categories are formed, the two
body is categorized as “poor,” “marginal,” “fair,” lowest ones were merged so that the water body
“good,” or “excellent” (CCME 2001). According to could be classified into five water quality categories,
CCME methodology, index calculation requires that and thus, be comparable with the indices applied in
at least four quality parameters sampled at minimum this study (Dinius 1987).
four times are used. However, the methodology does
not specify which or how many quality parameters Weighted Arithmetic WQI
should be included, and thus, it is up to the user to
select them. In the present work, the selection of the Weighted Arithmetic WQI classifies the water bod-
quality parameters for the calculation of CCME ies according to their suitability for drinking pur-
WQI as well as of the corresponding objective pose. Its computation requires the selection of the
values was based on a similar study undertaken in quality parameters, which will be included, as well
Polyphytos reservoir-Aliakmon River in Greece by as the definition of the corresponding boundary con-
Alexakis et al. (2016). The computation of CCME centration values. Through a sub-index function, the
index was implemented by applying the calculator concentration of each quality parameters is trans-
version 1.0, prepared by CCME (2001) for this formed into a unitless value, indicating the effect
purpose, available at http://www.mae.gov.nl. of the given parameter to the overall water quality.
ca/waterres/quality/background/cwqi.html (accessed Furthermore, a weight unit is attributed to each
on May 31, 2019). parameter reflecting its relative importance. Finally,
the overall WQI derives as the weighted arithmetic
Dinius’ Second Index mean of the individual sub-index values and ranges
from 0 to 100 (and above), indicating worse water
This WQI was developed by Dinius in 1987. It quality as the index numerical value increases. Ac-
evaluates the overall water quality taking into ac- cording to this score, the quality condition in the
count its suitability for six water uses: public water water body is categorized into five classes, namely
supply, recreation, fish, shellfish, agriculture, and “excellent,” “good,” “poor,” “very poor,” and “un-
industry. Dinius’ second index includes 12 quality suitable” for drinking purpose water quality (Tyagi
parameters, whose concentrations are transformed et al. 2013). In this study, the selection of the water
into unitless numerical values by applying the cor- quality parameters was based on a prior literature
responding sub-index functions. Then, the normal- search on the most frequently included parameters
ized sub-index values along with the respective when calculating Weighted Arithmetic WQI as well
weighting factors for each of the included parame- as on the availability of the necessary water quality
ters are combined in a multiplicative aggregation data. The boundary concentration values were spec-
function, through which the overall index is derived. ified synthetically in accordance with the guidelines
Dinius’ final index is a single numerical value of the Directive 98/83/EC, Joint Ministerial
505 Page 8 of 14 Environ Monit Assess (2019) 191: 505

Decision Υ2/2600/2001 (2001), Joint Ministerial between two parameters x and y is given by the follow-
Decision 46399/4352/86 (1986), World Health Or- ing equation (Boddy and Smith 2009):
ganization (1993), and Canadian Council of Minis-
nð∑xyÞ−ð∑xÞð∑yÞ
ters of the Environment (2001). r ¼ rhffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ih iffi ð1Þ
Each of the indices described above was calculated n∑x −ð∑xÞ n∑y −ð∑yÞ
2 2 2 2

separately per sampling (29 samplings were implement-

ed from August 2005 to November 2006 at each station) where n is the number of observations.
for each of the three gauging stations. In this way, a time It is important to clarify that the investigation of the
series of index scores was acquired for each location. most significant parameters when applying a specific
Then, these calculated index values were corresponded WQI is a quite common practice worldwide. Given that
to specific quality classes (from 1 to 5) according to the the majority of the most widely applied indices have
classification scheme of each methodology index. In the been developed in the USA and Canada, many re-
present work, in order to simplify the illustration of the searchers from different parts of the world have
results, for all the methodologies, quality class 5 was attempted to modify and amend them to fit the local
considered as the best quality status and class 1 as the conditions of specific regions. In most cases, WQIs can
worst one. also be applied and be effective using a different or a
Based on this classification, the variation of the qual- reduced number of the parameters initially proposed by
ity class over the monitoring period was determined for the author. This latter case provides numerous advan-
each methodology index (Fig. 2). Finally, Vosvozis tages in terms of the required sources since a smaller
River was classified in a certain quality category, per number of quality variables is translated into reduced
methodology, according to its overall water quality. For costs for the implementation of the corresponding sam-
this purpose, the worst quality scenario was taken into pling campaigns and laboratory analyses. In order that a
consideration, which means that, for each index applied, “minimum” yet representative index is developed, a
its lowest value (or its highest in case of an ascending prior determination of the most important parameters
scale index) recorded across the three gauging stations is needed. This exercise also shows which parameters
over the entire monitoring period was selected to clas- are more sensitive in determining the index score.
sify the river into a certain quality class according to its
overall quality. The results of the final classification of
Vosvozis River per methodology are presented in
Table 2. Results and discussion
The computation of the individual WQIs was follow-
ed by the demonstration of the “most important” water Figure 2 depicts the variation of the quality class, per
quality parameters per methodology. Practically, this methodology, at the three gauging stations along the
was implemented by determining Pearson’s correlation river. It is noted that the variation of the quality class
coefficient between each individual parameter and the was selected to be illustrated, instead of that of the
final value of each index applied. Correlation coefficient numerical value of the individual indices, in order to
r is a numerical measure of how strong is the linear achieve compatibility among the methodologies, since
relationship between two parameters, for which a data three of them, i.e., Prati’s Index, Oregon WQI and
set of observations is available. Correlation values range Weighted Arithmetic WQI, are expressed in different
from −1 to +1, with −1 indicating a perfect negative numerical scales. By analyzing the graphs of Fig. 2,
relationship between the parameters and +1 a perfect several observations can be made. First, it becomes clear
positive relationship. A positive relationship between that considerable differences arise in the qualitative
the two variables denotes that when the independent classification when applying a different WQI method-
variable increases, the dependent one increases as well, ology. This is mainly due to the quality parameters each
while a negative relationship denotes exactly the oppo- methodology utilizes, the form of the aggregation func-
site, i.e., increase in the values of one parameter causes tion, and the classification system, as well. In particular,
decrease in the value of the other. A value of zero (or it is observed that for the majority of the individual
near zero) means that the parameters being examined samplings at all three stations, the application of Dinius
tend to be unrelated. The correlation coefficient r WQI leads to the classification of the river in higher
Environ Monit Assess (2019) 191: 505 Page 9 of 14 505

Fig. 2 Variation of quality class 5

during the monitoring period in (a)

Quality class
4
Vosvozis river, based on the seven
applied WQIs at the three gauging 3
stations: a station located near 2
village Folia (GS1); b station lo-
1
cated near village Paradimi
(GS2); c station located upstream 0
and near Ismarida lagoon (GS3)

Prati's Index of Pollution Bhargava's Index Oregon WQI

NSF WQI CCME WQI Weighted Arithmetic WQI
Dinius WQI

5
(b)
4

Quality class
3

Prati's Index of Pollution Bhargava's Index Oregon WQI

NSF WQI CCME WQI Weighted Arithmetic WQI

Dinius WQI

5
(c)
4
Quality class

Prati's Index of Pollution Bhargava's Index Oregon WQI

NSF WQI CCME WQI Weighted Arithmetic WQI

Dinius WQI

quality classes, giving results ranging mostly between entire monitoring period at all three stations. Regarding
classes 3 and 5. Respectively, Bhargava’s and NSF Oregon and CCME WQIs, it can be seen that these
WQIs also give relatively high quality class results, indices are rather “strict” and less sensitive to changes
classifying the water body between classes 2 and 5 for in water quality compared to the rest of the methodolo-
the first case, and mainly between classes 3 and 4 for the gies, since they classify the water body mostly into the
second one. Slightly lower classification results derive 1st class and between classes 1 and 2, respectively,
when applying Prati’s Index, which categorizes presenting much lower variability. Finally, with regards
Vosvozis River between classes 1 and 4. Actually, it is to the Weighted Arithmetic WQI, its application clas-
important to note that Prati’s Index gives results of sifies the river almost steadily into the lowest quality
exclusively lower or equal quality class, compared to class (1st class), which appears reasonable given the
the two previous indices, Bhargava and NSF, for the nature of this index. The findings of this research
505 Page 10 of 14 Environ Monit Assess (2019) 191: 505

Table 2 Final classification results, based on the worst quality scenario, for each WQI applied

WQI Quality class into which the river was Quality characterization Gauging station at which the worst quality
categorized (1–5) was recorded

Prati’s Index of 1 Heavy pollution GS3

pollution
Bhargava’s Index 2 Poor water quality GS2
Oregon WQI 1 Very poor water quality GS2
NSF WQI 3 Medium water quality GS2
CCME WQI 1 Poor water quality GS2
Dinius Index 1 Unacceptable water quality GS3
Weighted 1 Unacceptable for drinking GS2
Arithmetic WQI purposes

regarding the comparative effectiveness of the individ- classification results compared to CCME. Similar find-
ual WQIs in evaluating and classifying a water body are ings regarding the CCME-NSF relative performance
in good agreement with our previous study (Zotou et al. emerged from a research conducted by Finotti et al.
2018) where the comparative performance of the same (2015). However, it should be noted that only two of
indices in a lacustrine water body in Greece, namely, the aforementioned studies, i.e., those of Hamlat et al.
Polyphytos Reservoir, was examined. Nevertheless, on- (2017) and Akkoyunlu and Akiner (2012), refer to
ly Dinius and Weighted Arithmetic WQIs showed a Mediterranean water bodies, but they have been imple-
different behavior between the two case studies, which, mented in northern (Algeria) and eastern (Turkey) Med-
in the case of Dinius WQI, was probably due to the iterranean region where the conditions are different.
consideration of a different water use (“potable water With respect to the relative quality condition of water
supply” in the case of Polyphytos Reservoir and “fish at the three gauging stations, it was observed that, inde-
life” in the current study). pendently of the methodology applied, station GS1 pre-
To date few researchers have attempted to undertake sents higher water quality compared to the downstream
comparative evaluation of WQIs encompassing a sig- station GS2, almost for the entire monitoring period,
nificant number of them. Most studies have only fo- which is clearly explained by the location of the latter
cused on comparing a limited number of indices, with at a near distance from the agricultural areas of the
CCME, NSF, and Oregon being the most usually watershed and next to the wastewater treatment plant
applied. Therefore, given that none of the existing of the city of Komotini. Furthermore, it was concluded
studies until now examines exactly the same that the water quality status of the further downstream
methodologies as those applied in the present work, it station GS3 over the monitoring period is characterized
is hard to reliably and comprehensively confirm the as slightly higher or equal to that of the upstream station
current findings against previous results. However, GS2, which probably indicates the partial assimilation of
studies such as those of Darvishi et al. (2016) and the polluting loads along the part of the stream between
Hamlat et al. (2017) give similar results to the present the two stations. It should be mentioned that the above
study, regarding the relative performance of CCME, conclusions concerning the spatial changes in water
NSF, and Oregon WQIs, and verify the strict and less quality along the river can be more easily detected when
sensitive nature of the latter. This has also been examining the numerical values of the individual WQIs
demonstrated by Akkoyunlu and Akiner (2012) who (instead of the corresponding quality classes), which are
examined these three indices concluding that NSF not included in this paper, as mentioned, for reasons of
WQI leads to the highest qualitative classification illustration (Zotou 2017).
whereas Oregon to the lowest one. This was partially Table 2 summarizes the final results of the quali-
supported by Hashim et al. (2015), who found that tative classification of Vosvozis River, as derived
Oregon gives the highest WQI values among the three from each methodology. This classification repre-
indices, agreeing, however, in that NSF produces higher sents the overall water quality of the water body
Environ Monit Assess (2019) 191: 505 Page 11 of 14 505

and is based on the worst quality scenario, namely, it could be safely claimed is that the “important param-
the worst quality condition recorded at the three eters” as determined in this study concern the applica-
stations over the entire monitoring period. It is ob- tion of a particular index in a particular case. This means
served that Vosvozis River is classified into the 3rd that a specific WQI methodology may be safely applied
quality class according to NSF WQI, one class lower in the examined water body, by incorporating only the
(2nd class) according to Bhargava’s Index, while for “important variables,” and be adequately effective. Par-
the rest of the methodologies, it is categorized into ticular attention should be given in generalizing these
the lowest quality class (class 1). results, since further investigation in more case studies is
Table 3 depicts the correlation coefficients between required to examine whether the important parameters
water quality parameters and the final WQI value, for remain the same when applying the same WQI in a
each methodology applied. Parameters presenting a different water body.
stronger linear relationship (higher correlation coeffi-
cient) with the computed results are shown in italics.
However, the correlation coefficient value between each Conclusions
individual parameter and the final index result, when
applying a certain methodology, depends on a variety of In this paper, an attempt was made to apply and inves-
factors, such as the combination of the quality parame- tigate the comparative performance of seven widely
ters included in the index, the way in which the param- used WQIs in a fluvial surface water body, namely,
eter is used within the computational framework, and Vosvozis River in North Greece. In particular, the indi-
the sample itself, namely, the concentration levels in ces examined were the Prati’s Index of Pollution,
which this parameter is recorded and how much these Bhargava WQI, Oregon WQI, NSF WQI, CCME
levels may affect the final quality result. Consequently, WQI, Dinius second index, and the Weighted Arithmet-
the results of the “important parameters” are driven both ic WQI. WQIs were proved to be a particularly promis-
from the “data set” and the “methodological frame- ing tool in the direction of the qualitative assessment and
work.” The extent to which each of these two compo- classification of surface water bodies. Specifically, giv-
nents contributes is not completely clear. However, what en the advantages arising from their application, such as

Table 3 Correlation coefficients between water quality parameters and each WQI methodology (parameters presenting a stronger linear
relationship with the computed results, i.e., higher correlation coefficient, are presented in italics)

Parameter WQI

Prati’s Index Bhargava’s WQI Oregon WQI NSF WQI CCME WQI Dinius WQI Weighted
Arithmetic
WQI

T – − 0.481 − 0.156 − 0.166 – – –

DO − 0.379 0.472 0.261 0.585 0.439 0.502 − 0.135
pH 0.067 – − 0.022 − 0.014 0.045 0.064 − 0.291
EC – − 0.634 – – – − 0.559 0.040
BOD5 0.876 − 0.773 − 0.410 − 0.663 − 0.474 − 0.655 − 0.011
COD 0.869 – – – – – − 0.173
NO2−-N – – – – – – 0.466
NO3−-N − 0.165 – – − 0.048 – − 0.246 0.044
NH4+-N − 0.038 0.076 – – − 0.283 – 0.886
(NO2−-N) + (NO3−-N) – – – – − 0.130 – –
(NO3−-N) + (NH4+-N) – – − 0.292 – – – –
TKN – – – – – – 0.197
TP – – − 0.022 − 0.464 − 0.498 – 0.517
505 Page 12 of 14 Environ Monit Assess (2019) 191: 505

the establishment of a common reference framework for our previous study (Zotou et al. 2018) where exactly
the comparison of different water bodies or the oppor- the same indices had been examined. Additionally,
tunity they provide to detect changes in water quality the findings emerged from the current work are to a
over a certain period or across geographic areas, WQIs great extent supported by similar studies examining a
could play a key role in the implementation of the smaller number of indices (Darvishi et al. 2016;
European Water Framework Directive. Through their Hamlat et al. 2017; Akkoyunlu and Akiner 2012;
application, the different member States could success- Hashim et al. 2015; Finotti et al. 2015). The compar-
fully assess as well as monitor the evolution in the water ison with these studies revealed analogous results
quality status of their aquatic systems in order to meet regarding the relative performance of NSF, CCME,
the requirements of the Directive. It should be men- and Oregon WQIs.
tioned, though, that according to the WFD, Greece The classification results produced by this research
belongs to the Mediterranean Geographical Intercalibra- indicate that, among all applied indices, CCME WQI
tion Group (GIG) along with Spain, Italy, Cyprus, appears to be more appropriate, since it is both “conser-
France, Portugal, and Slovenia. Under this context and vative” enough to avoid water quality overestimation
assuming water systems of similar bio-geophysical and more “sensitive” compared to Oregon WQI, so that
characteristics within each GIG, specific indices for the temporal or spatial changes can be reflected.
the evaluation of the ecological quality have been Finally, given that the above results were based on
established, aiming at ensuring comparability among limited case studies, further investigation is needed in
the states of each eco-region. It becomes clear that, order to draw more reliable conclusions. The implemen-
although “type specific” approach is crucial in the eval- tation of the individual methodologies in an adequate
uation of water bodies, since it allows for the consider- number of water bodies characterized by different hy-
ation of the “natural conditions,” a rather unifying con- drological and climatic conditions is required before
text such as that proposed by the WFD is indispensable. applying them on a wide level.
However, the indices applied within the Mediterranean
GIG utilize solely biotic variables to assess the ecolog-
ical status. Therefore, a comparison of some of the References
applied WQIs, which are predominately based on phys-
icochemical parameters, against the methodologies pro- Abbasi, T., & Abbasi, S. A. (2012). Water Quality Indices. 1st Ed.,
posed by the WFD, would be extremely useful. Elsevier, New York, NY, USA, ISBN 9780444543059, 384p.
Adimalla, N., Li, P., & Venkatayogi, S. (2018). Hydrogeochemical
It becomes clear, though, that utilizing WQIs on a
evaluation of groundwater quality for drinking and irrigation
wider scale, e.g., European or Mediterranean, requires purposes and integrated interpretation with Water Quality
the prior testing of them under different local conditions, Index studies. Environmental Processes, 5(2), 363–383.
in order to ensure that they behave in the same way, and Akkoyunlu, A., & Akiner, M. E. (2012). Pollution evaluation in
streams using water quality indices: a case study from
thus, to be able to suggest one or more. Besides, the
Turkey's Sapanca Lake Basin. Ecological Indicators, 18,
suggested indices should at a later stage be amended to 501–511.
fit the WFD requirements as well as to be better adapted Akoteyon, I. S., Omotayo, A. O., Soladoye, O., & Olaoye, H. O.
to the natural conditions of each region. (2011). Determination of water quality index and suitability
The results derived by the present work led to the of urban river for municipal water supply in Lagos–Nigeria.
European Journal of Scientific Research, 54(2), 263–271.
conclusion that considerable differences arise in the Alexakis, D., Tsihrintzis, V. A., Tsakiris, G., & Gikas, G. D.
qualitative classification when applying a different (2016). Suitability of Water Quality Indices for application
WQI. More specifically, it was found that Bhargava in lakes in the Mediterranean. Water Resources Management,
and NSF WQIs tend to classify water bodies into 30(5), 1621–1633.
Almeida, C., Quintar, S., González, P., & Mallea, M. (2008).
superior quality classes compared to Prati, Oregon, Assessment of irrigation water quality. A proposal of a qual-
and CCME WQIs. Respectively, slightly lower qual- ity profile. Environmental Monitoring and Assessment,
ity class results derive when applying Prati’s Index, 142(1–3), 149–152.
whereas Oregon and CCME WQIs are proved to be Al–Mutairi, N., Abahussain, A., El–Battay, A. (2014). Application
of water quality index to assess the environmental quality of
“stricter” and less “flexible” methodologies, giving Kuwait Bay. In International Conference on Advances in
results ranging steadily between the lowest quality Agriculture, Biological & Environmental Sciences (AABES
classes. These results were found to be in line with 2014), pp 15–16.
Environ Monit Assess (2019) 191: 505 Page 13 of 14 505

Alobaidy, A. H. M. J., Abid, H. S., & Maulood, B. K. (2010). using water quality indices. Journal of Water Reuse and
Application of water quality index for assessment of Dokan Desalination, 7(2), 228–245.
lake ecosystem, Kurdistan region, Iraq. Journal of Water Haque, M. M., Kader, F., Kuruppu, U., Rahman, A. (2015).
Resource and Protection, 2(09), 792–798. Assessment of water quality in Hawkesbury–Nepean River
Bhargava, D. S. (1983). Use of water quality index for river in Sydney using Water Quality Index and multivariate anal-
classification and zoning of Ganga River. Environmental ysis. In Proceedings of the 21st International Congress on
Pollution Series B, Chemical and Physical, 6(1), 51–67. Modelling and Simulation, pp 2493–2499.
Boddy, R., & Smith, G. (2009). Statistical methods in practice: for Hashim, S., Yuebo, X., & Hashim, I. (2015). Comparative indices
scientists and technologists. John Wiley & Sons, Inc., evaluation to restore urban rivers water quality by using
Chichester, UK, ISBN 978–0–470–74664–6, 246p. bacterial technology. Kasmera Journal, 43, 198–209.
Boskidis, I., Gikas, G. D., Pisinaras, V., & Tsihrintzis, V. A. Horton, R. K. (1965). An index number system for rating water
(2010). Spatial and temporal changes of water quality, and quality. Journal of Water Pollution Control Federation, 37,
SWAT modeling of Vosvozis river basin, North Greece. 300–305.
Journal of Environmental Science and Health Part A, Joint Ministerial Decision 46399/4352/86 (1986) “Quality re-
45(11), 1421–1440. quired for surface water intended for the abstraction of
Brown, R. M., McClelland, N. I., Deininger, R. A., & Tozer, R. G. drinking water, bathing, fish life in freshwater and shellfish,
(1970). A water quality index–do we dare? Water Sew Works, methods of measurement and frequency of sampling and
117, 339–343. analysis of surface water intended for the abstraction of
Canadian Council of Ministers of the Environment (2001). drinking water, in compliance with Council Directives 75/
Canadian water quality guidelines for the protection of aquat- 440/EEC, 76/160/EEC, 78/659/EEC, 79/923/EEC and 79/
ic life: CCME Water Quality Index 1.0, User’s Manual. In 869/EEC”. Athens, Greece: National Printing House of
Canadian environmental quality guidelines, Canadian Greece (in Greek).
Council of Ministers of the Environment, Publication No. Joint Ministerial Decision Υ2/2600/2001 (n.d.). “Quality of
1299, Winnipeg, Canada, ISBN 1–896997–34–1, water intended for human consumption in compliance
http://ceqg-rcqe.ccme.ca/download/en/138. Accessed 1 with Council Directive 98/83/EC of 3 November 1998”
June 2019. Athens, Greece: National Printing House of Greece (in
Chauhan, A., & Singh, S. (2010). Evaluation of Ganga water for Greek).
drinking purpose by water quality index at Rishikesh,
Kumari, S., & Rani, J. (2014). Assessment of Water Quality Index
Uttarakhand, India. Report and Opinion, 2(9), 53–61.
of ground water in Smalkhan, Haryana. International
Cude, C. G. (2001). Oregon water quality index: a tool for eval-
Journal of Latest Research in Science and Technology,
uating water quality management effectiveness. Journal of
3(6), 169–172.
the American Water Resources Association, 37(1), 125–137.
Mitra, P., & Reddy, P. B. (2016). Application of Water Quality
Darvishi, G., Kootenaei, F. G., Ramezani, M., Lotfi, E., &
Index (WQI) as a tool for assessment of pollution status of
Asgharnia, H. (2016). Comparative investigation of river
Shivna River at Mandsaur, MP India. Trends in Life Science,
water quality by OWQI, NSFWQI and Wilcox Indices (case
5, 4–11.
study: the Talar River–Iran). Archives of Environmental
Moyel, M. S., & Hussain, N. A. (2015). Water quality assessment
Protection, 42(1), 41–48.
of the Shatt al–Arab River, Southern Iraq. Journal of Coast
Debels, P., Figueroa, R., Urrutia, R., Barra, R., & Niell, X. (2005).
Life Medicine, 3(6), 459–465.
Evaluation of water quality in the Chillan River (Central
Chile) using physicochemical parameters and a modified Pesce, S. F., & Wunderlin, A. (2000). Use of Water Quality Indices
Water Quality Index. Environmental Monitoring and to verify the impact of Cordoba city (Argentina) on Suquia
Assessment, 110(1–3), 301–322. River. Water Research, 34(11), 2915–2926.
Deshmukh, K. K., & Aher, S. P. (2016). Assessment of the impact Pisinaras, V., Petalas, C., Tsihrintzis, V. A., & Zagana, E. (2007). A
of municipal solid waste on groundwater quality near the groundwater flow model for water resources management in
Sangamner City using GIS approach. Water Resources the Ismarida plain, North Greece. Environmental Modeling
Management, 30(7), 2425–2443. and Assessment, 12(2), 75–89.
Dinius, S. H. (1987). Design of an index of water quality. Journal Prati, L., Pavanello, R., & Pesarin, F. (1971). Assessment of
of the American Water Resources Association, 23(5), 833– surface water quality by a single index of pollution. Water
843. Research, 5(9), 741–751.
Ewaid, S. H. (2016). Water quality assessment of Al–Gharraf Raju, A., & Singh, A. (2017). Assessment of groundwater quality
River, south of Iraq by Canadian Water Quality Index and mapping human health risk in Central Ganga Alluvial
(CCME WQI). Iraqi Journal of Science, 57(2A), 878–885. Plain, Northern India. Environmental Processes, 4(2), 375–
Finotti, A. R., Finkler, R., Susin, N., & Schneider, V. E. (2015). 397.
Use of water quality index as a tool for urban water resources Shah, K. A., & Joshi, G. S. (2017). Evaluation of water quality
management. International Journal of Sustainable index for River Sabarmati, Gujarat, India. Applied Water
Development and Planning, 10(6), 781–794. Science, 7(3), 1349–1358.
Gikas, G. D. (2017). Water quantity and hydrochemical quality of Trikoilidou, E., Samiotis, G., Tsikritzis, L., Kevrekidis, T., &
Laspias River, North Greece. Journal of Environmental Amanatidou, E. (2017). Evaluation of water quality in-
Science and Health Part A, 52(14), 1312–1321. dices adequacy in characterizing the physico–chemical
Hamlat, A., Guidoum, A., & Koulala, I. (2017). Status and trends water quality of lakes. Environmental Processes, 4(1),
of water quality in the Tafna catchment: a comparative study 35–46.
505 Page 14 of 14 Environ Monit Assess (2019) 191: 505

Tyagi, S., Sharma, B., Singh, P., & Dobhal, R. (2013). Water Yogendra, K., & Puttaiah, E. T. (2008). Determination of
quality assessment in terms of Water Quality Index. water quality index and suitability of an urban water
American Journal of Water Resources, 1(3), 34–38. body in Shimoga Town, Karnataka. In Proceedings of
World Health Organization. (1993). Guidelines for drinking water Taal 2007: The 12th world lake conference, pp 342–346.
quality, Vol. 1: recommendations (2nd ed.p. 150). Geneva: Zotou, I. (2017). Comparative assessment of Seven Water Quality
World Health Organization. Indices (WQIs) based on measurements in surface water
Xia, J., Xu, G., Guo, P., Peng, H., Zhang, X., Wang, Y., & Zhang, systems of Greece. MSc Thesis, National Technical
W. (2018). Tempo–spatial analysis of Water Quality in the University of Athens (in Greek).
Three Gorges Reservoir, China, after its 175–m experimental Zotou, I., Tsihrintzis, V. A., Gikas, G. D. (2018). Comparative
impoundment. Water Resources Management, 32(9), 2937– assessment of various water quality indices (WQIs) in
2954. Polyphytos Reservoir–Aliakmon River, Greece. In MDPI
Yaseen, Z. M., Ramal, M. M., Diop, L., Jaafar, O., Demir, V., & Proceedings, 2(11), p. 611, DOI: https://doi.org/10.3390
Kisi, O. (2018). Hybrid adaptive neuro–fuzzy models for /proceedings2110611.
Water Quality Index estimation. Water Resources
Management, 32(7), 2227–2245.
Publisher’s note Springer Nature remains neutral with regard to
Yisa, J., & Jimoh, T. (2010). Analytical studies on water quality
jurisdictional claims in published maps and institutional
index of river Landzu. American Journal of Applied
affiliations.
Sciences, 7(4), 453–458.