Sei sulla pagina 1di 10

Ecological Indicators 78 (2017) 243–252

Contents lists available at ScienceDirect

Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind

Original Articles

Groundwater quality classification derivation using


Multi-Criteria-Decision-Making techniques
Sina Zahedi a,∗ , Ali Azarnivand b , Nastaran Chitsaz c
a
Department of Renewable Energies and Environment, Faculty of New Sciences and Technologies, University of Tehran, Tehran, 31587-77871, Iran
b
Department of Irrigation & Reclamation Engineering, Faculty of Agricultural Engineering & Technology, College of Agriculture & Natural Resources,
University of Tehran, Karaj, Tehran, Iran
c
University of Tehran, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: The paper introduces a novel framework of evaluating and validating the groundwater quality index
Received 26 November 2016 (GWQI) by employing two Multi-Criteria-Decision-Making models (MCDMs) such as Technique for Order
Received in revised form 3 March 2017 of Preference by Similarity to Ideal Solution (TOPSIS) and Compromise Programming (CP). GWQIs are
Accepted 7 March 2017
plagued by discrete values which are attributed to excellent, good, poor, very poor, and unsuitable for
drinking purpose classes. Furthermore, the existence of chemical parameters with low weights but high
Keywords:
concentration values (or vice versa) reduces equilibrium of evaluations. Based on the literature review,
Groundwater
MCDMs have not been applied so far to validate groundwater quality indexes. This simple yet practical
Multi Criteria Decision Making
Water Quality Index (WQI)
structure assists the analysts to investigate robustness of a water quality index. In order to validate the
Arid region classes, the intervals of classic GWQI were modified to figure out what changes were made to the classes of
Drinking purpose samples in the ranking table considering quality of 92 wells in the Varamin plain, Iran. After employment
of the MCDMs, the results revealed that WQI could rarely evaluate water quality classes accurately.
Several wells which were classified as “excellent quality” were belonged to the lower quality class. It also
came true for other classes. The results demonstrated that TOPSIS could present a more precise analysis
for the classes with a low number of wells. Moreover, increases in value of the sensitivity parameter of
CP technique reduced the accuracy of results. Owing to proper application of MCDMs, one can suggest
the use and development of MCDM-based water quality assessment models for future researches.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction level besides the groundwater quality degradation (Srinivas et al.,


2015). Therefore, in order to meet the requirements in a broad
Groundwater is one of the major existing water resources area of applications, quality of the groundwater resources is as
and is being used by a large number of the world population important as its quantity, and should be essentially taken into
for their basic needs such as drinking, agricultural activities and a full consideration (Aghazadeh and Asghari-Mogaddam, 2010).
industries specifically in arid and semiarid regions (Wu and Sun, The quality of recharged water, atmospheric precipitation, inland
2016; Li et al., 2016b; Chitsaz and Azarnivand, 2016). In gen- surface water, and sub-surface geochemical processes affect the
eral, population growth accompanied by the ascending trend of groundwater quality (Vasanthavigar et al., 2010). Consequently,
urbanization, agriculture and industrialization in the developing groundwater contamination threatens the human health, eco-
countries have imposed unsustainable demands on the natu- nomic development and social prosperity (Milovanovic, 2007). In
ral aquifer systems (Krishan et al., 2016; Howard and Howard, Iran, Groundwater plays a central role for drinking purposes, and
2016; Li et al., 2015). In addition, over-exploitation and indiscreet satisfying agricultural, domestic and industrial water demands
use of groundwater has influenced the groundwater availability (Sadat-Noori et al., 2013). An imprudent exploitation of groundwa-
and quality, having led to a noticeable decline of groundwater ter resources accompanied by consecutive droughts in recent years
has degraded the quality of groundwater and decreased ground-
water level of aquifers in Iran (Hosseinifar and Mirzaei Aminiyan,
2015). For instance, results of a research in Kashan province of Iran
∗ Corresponding author. revealed that just 47% of water samples were unsuitable for irri-
E-mail addresses: Zahedi.Sina@ut.ac.ir (S. Zahedi), Azarnivand ali@ut.ac.ir gation (Jamshidzadeh and Mirbagheri, 2010). Throughout another
(A. Azarnivand), Nastaranchitsaz@ut.ac.ir (N. Chitsaz).

http://dx.doi.org/10.1016/j.ecolind.2017.03.015
1470-160X/© 2017 Elsevier Ltd. All rights reserved.
244 S. Zahedi et al. / Ecological Indicators 78 (2017) 243–252

research regarding groundwater quality assessment of Amol-Babol (2013a,c), TOPSIS performed properly in sensitivity analysis of dif-
Plain, Iran, the thematic maps of salinity hazard and Na+ percent- ferent physiochemical parameters’ weights.
age showed that the water quality deteriorated gradually across According to the literature review, various improvements and
the plain (Narany et al., 2015). applications have been be associated with MCDMs in uncertainty
Several methods have already been developed, in order to pro- analysis, sensitivity analysis and prioritization mechanism. How-
vide adequate groundwater quality indicators for different regions. ever, the current paper uses MCDMs to answer the following
Water Quality Index (WQI), developed by Brown et al. (1970), is questions: (1) Are the classifications of WQI accurate? (2) Can dif-
one of the most commonly used methods for assessment and man- ferent assumptions of the MCDMs affect the ranking of wells in each
agement of groundwater quality (Hosseini-Moghari et al., 2015; class? (3) Are MCDMs sensitive to numbers of wells in each water
Mosaferi et al., 2015; Annapoorna and Janardhana, 2015; Sadat- quality class?
Noori et al., 2013; Rizwan and Gurdeep, 2010; Saeedi et al., 2010;
Backman et al., 1998).
On the other hand, many investigators believe that this method 2. Materials and methods
has some structural drawbacks preventing it to achieve desirable
results. WQI is evaluated based on summation of relative weights of 2.1. Case study
each chemical parameter. As a result, a well could possibly present
a higher WQI value just because of considerable amount of a con- Located in south east of Tehran province, Varamin plain lies
centration in one chemical parameter which has a high weight. On between 51◦ 23’ to 51◦ 54’ longitude and 35◦ 07’ to 35◦ 32’ lati-
the other hand, a well with lower concentration in that high-weigh tude in the center of Iran (Fig. 1a). The area of this plain is about
parameter but higher values in other parameters might present a 1042.3 Km2 while its mean altitude is about 808–1161 m above
lower WQI. Moreover, changes in significance of each parameter mean sea level (AMSL). The climate of the area is arid and semi-arid,
would adversely affect the results of classifications considerably considering Dommartin and Ambreje categories, respectively, with
which can be associated with unrealistic relative weights. In this an annual precipitation amount equal to 157.5 mm. The Precipita-
regard, Lermontov et al. (2009) described that utilizing the classic tion data obtained from meteorological stations demonstrated that
WQI could lead to create deterministic and inflexible classifica- rainfall generally occurs from October to May with maximum and
tions related to inaccuracies in knowledge and data. In line with minimum rate in March and October, respectively. Moreover, the
the above issue, Dahiya et al. (2007) elucidated that the final score mean annual temperature is about 16.4 ◦ C and also the minimum
of WQI could be dramatically influenced without a valid scientific and maximum mean monthly temperatures are measured to be
justification by some parameters in the index equations. about 1.7 ◦ C in January and 30.1 ◦ C in July, respectively. According
With this knowledge in hand, an effective tool is needed to mit- to the Land-use map of the plain, agriculture is the main industry
igate the aforementioned shortcomings. Some investigators have in Varamin (Fig. 1b).
used methods such as Schoeller diagram. However, because of high Tehran Regional Water Authority (TRWA) is responsible for
sensitivity to the values of each parameter, this technique might monitoring wells related to water quality. In this region 92 wells
lead to generation of unreality-based outcomes. In Schoeller dia- are being monitored every six months in order to measure water
gram method, violation of only one parameter from the proposed quality from 2002 to 2014. In order to analyze groundwater quality
standard is expected to penalize all the parameters. Hence, devel- of the aquifer, the samples were transferred to central laboratory of
opment of a tool for reliable and flexible decision-making in water TRWA. The available parameters included T.H, S.A.R, K, Na, Mg, Ca,
quality assessment seems to be vital (Nikolić et al., 2010). Some HCO3 , Cl, SO4 , pH, T.D.S, E.C, and NO3 (Table 1). In addition, obser-
attempts have been made to improve efficiency of WQI in the liter- vation wells are monitored every month to evaluate groundwater
ature by improving the weight’s calculation and considering fuzzy table and aquifer loss. According to Fig. 1(c), groundwater level has
boundaries. Li et al. (2010) integrated WQI and entropy weight decreased to 35 m from 1990 to 2014. This situation revealed the
to improve the objectivity of the approach, and their results indi- occurrence of a marked over-exploitation in the region.
cated that the entropy-weighted WQI could make the assessment
results more objective and reasonable. Li et al. (2014) further inte- 2.2. Water quality index
grated fuzzy theory into the entropy-weighted WQI to treat the
uncertainty associated with water quality classification. Water Quality Index (WQI) is defined as a technique of rating to
The current research applies Multi Criteria Decision Making prepare the composite effect of individual water quality parameter
(MCDM) techniques to check validation of WQI classes which has on the overall quality of water. Moreover, it can be calculated based
not captured the investigators attention. Two MCDMs, namely, on the viewpoints of human water consumption. Indeed, evaluation
compromise programming (CP), and technique for order preference of groundwater quality for drinking determines its suitability for
by similarity to ideal solution (TOPSIS) with different mathemat- different purposes based upon the specific standards set by various
ical assumptions are used in the current study. MCDMs have agencies including the drinking water standards of World Health
shown efficient performance in such fields as agricultural water Organization (WHO, 2004). In this method, the weights of various
quality assessment (Hernandez and Uddameri, 2010), water qual- water quality parameters are assumed to be inversely proportional
ity monitoring (Khan and Maity, 2016; Chang and Lin, 2014; Li to the recommended standards for the corresponding parameters.
et al., 2012), natural hazard management (Chitsaz and Banihabib, According to their influence on primary health, a weight factor
2015), health risk assessment (Li et al., 2016a), reservoir operation should be determined for each of the ten parameters, including
(Bozorg-Haddad et al., 2016a,b), soil erosion alleviation (Chitsaz K+ , Na+ , Ca+ , Mg+ , SO4 + , HCO− 3 , Cl− , NO− 3 , TDS, and pH.
and Malekian, 2016), biodiversity rehabilitation (Azarnivand and The relative weight factors can be calculated for all parameters
Banihabib, 2016). Ning and Chang (2005) suggested CP as a robust using Eq. (1), according to their perceived weights in Table 2.
tool because of its appropriate application in screening the relo-
cation strategies of water quality monitoring stations. A combined wi
Wi = n (1)
application of TOPSIS method and entropy weight along with utiliz- i=1
wi
ing rough set theory showed flexibility of this technique in water
quality monitoring (Li et al., 2011, 2013b). According to Li et al. where (wi ) and (Wi ) are respectively the weight and relative weight
of each chemical parameter.
S. Zahedi et al. / Ecological Indicators 78 (2017) 243–252 245

Fig. 1. (A): Location of the studied area and the observation wells, (B): The land-use map of the studied area, (C): Average annual groundwater level of the study area during
1990–2013.

Table 1
The location and the average amount of each chemical parameter among years 2002–2014.

Wells Codes num1 num2 num3 num4 ... ... ... num90 num91 num92

UTM (X) 548261 548666 549500 549900 ... ... ... 577600 578500 579300
UTM (Y) 3912390 3915434 3901150 3913100 ... ... ... 3915800 3919000 3918800
Wilcox c4-s3 c4-s3 c4-s2 c4-s4 ... ... ... c3-s1 c3-s1 c2-s1
T.Ha (mg/L) 1800 2210 1025 545 ... ... ... 212 283 268
S.A.Rb 9.43 10.1 7.61 2.36 ... ... ... 3.54 3.56 0.70
T.D.Sc (mg/L) 4720 5471 3230 585 ... ... ... 687 803 400
pHd 8.02 7.37 7.18 7.83 ... ... ... 8.07 7.96 7.80
E.Ce (␮s/m) 6760 7930 4750 1590 ... ... ... 1025 1178 650
Cations (mEq/L) 76.2 92 45.1 16.4 ... ... ... 9.49 11.76 6.54
K (mg/L) 7.04 9.77 10.6 1.95 ... ... ... 4.30 4.69 1.17
Na (mg/L) 920 1092 560 126 ... ... ... 118 137.48 26.44
Mg (mg/L) 293 316 134 53.5 ... ... ... 23.2 28.44 19.57
Ca (mg/L) 239 365 190 130 ... ... ... 46.5 66.53 75.15
Anions (mEq/L) 74.2 93.2 45.8 15.9 ... ... ... 9.41 11.91 6.90
HCO3 (mg/L) 444 422 260 325 ... ... ... 216 209.90 161.69
Cl (mg/L) 752 1461 535 255 ... ... ... 84.02 146.42 39
SO4 (mg/L) 2181 2151 1260 149 ... ... ... 163.31 203.65 136.89
NO3 (mg/L) 0.30 0.30 0.20 0.30 ... ... ... 0.10 0.10 0.30
a
Total hardness.
b
Sodium adsorption ratio.
c
Total dissolved solids.
d
Potential of hydrogen.
e
Electrical conductivity.

The quality rating of each parameter is calculated using Eq. (2), Then, the sub quality index of each parameter (SIi ) should be cal-
in which concentration of each chemical parameter (Ci ) is divided culated by multiplication of their specific relative weights to their
by their WHO standard values (Si ), as shown in Table 2. quality rating scale, as referred in Eq. (3).

Ci
qi = × 100 (2)
Si SIi = qi × Wi (3)
246 S. Zahedi et al. / Ecological Indicators 78 (2017) 243–252

Fig. 2. Spatial distribution map of the WQI classification using Kriging method.

Table 2 Table 3
The weight (wi ) and relative weight (Wi ) of each chemical parameter. Classification of groundwater quality according to WQI.

Parameter WHO Standards (mg/L) Weight (wi ) Relative weights (Wi ) WQI range Type of Groundwater

K 12 2 0.056 <50 Excellent water


Na 200 4 0.111 50–99.99 Good water
Mg 50 3 0.083 100–199.99 Poor Water
Ca 75 3 0.083 200–299.99 Very poor water
HCO3 120 1 0.028 ≥300 Unsuitable for drinking purpose
Cl 250 5 0.139
SO4 250 5 0.139
pH 8.5 3 0.083 lyzing the spatial distribution of groundwater quality index. Spatial
T.D.S 500 5 0.139
analyst extension of Arc Map 9.3 was applied to reveal the spatial-
NO3 45 5
 
0.139
temporal condition of the groundwater quality parameters. Kriging
wi = 36 Wi = 1
method has also been employed so as to evaluate groundwater
quality conditions of the study area. The details of GIS application
and Kriging method can be found in Goovaerts (1997), ESRI (1999),
Finally, the water quality index of each sample can be computed
and the Supplementary material.
by summing all sub-indices, as defined in Eq. (4).

WQI = SIi (4) 2.3. Modified classification for water quality index

Subsequently, computed WQI value of each sample is specified The MCDMs such as TOPSIS and CP are applied to validate the
to determine the class of groundwater quality. As mentioned in WQI results. Employment of MCDMs may affect the ranks of wells
Table 3, the groundwater can be classified into the five categories and cause some fluctuations. For instance, if the changes occur
of excellent, good, poor, very poor, and unsuitable for human con- in boundaries of two classes, a well in lower class of WQI may
sumption. gain a better rank in MCDM models and even ascend to a better
Table 3 The classifications and water quality status must be classification. Therefore, the wells’ ranks should be assessed with
visualized by maps. In this paper, ARC GIS 9.3 was employed for ana- consideration of overlaps in different classes. The new classification
S. Zahedi et al. / Ecological Indicators 78 (2017) 243–252 247

Fig. 3. The WQI Values of used wells for classes: (a) Excellent water (0–50), (b) Good water (50,100), (c) Poor Water (100, 200), (d) Very poor water (200, 300), (e) Unsuitable
for drinking purpose (300 ≤ WQI).

of WQI method is as follows: WQI < 75 is considered as excellent classification, 51 wells are obtained as excellent water quality, 49
quality while WQI > 300 is unsuitable for drinking. The ranges of wells have good water quality, eight wells have poor water, eight
50 < WQI < 150, 100 < WQI < 250 and 200 < WQI < 350 are classified wells have very poor water and water of 16 wells are considered
as good, poor and very poor water quality, respectively. In this new unfit for drinking. Some wells sit in similar classes. The overlaps in
248 S. Zahedi et al. / Ecological Indicators 78 (2017) 243–252

Fig. 4. Final rank of each well using WQI, CP (p = 1), CP (p = 2), CP (p = ∞) and TOPSIS for classes (a) Excellent water, (b) Good water (c), Poor Water, (d) Very poor water.

ranges help us to investigate actual class of wells whose WQI values The measurement of the distance is based upon the p parameter,
are close to the boundary values. where Zeleny (1973) suggested 1, 2 and ∞ for p. The Eq. (6) is cor-
responded to Block distance and Euclidean distance, respectively.
2.4. Multi criteria decision making methods Furthermore, employment of ∞ for p in Eq. (6) is corresponded to
the Tchebycheff distance as follows (Pomerol and Barba-Romero,
Three basic elements involving criteria, alternatives and per- 2000):
formances of each alternative under each criterion constitute the ⎛ ⎞
structure of an MCDM problem (Madani and Lund, 2011). Two tech-
⎜ ⎟
niques involving TOPSIS and CP were used in this study as follows: ⎜ Wj ∗
fj − fij⎟
Assuming a decision matrix involving m wells and n chemical Lp (Welli ) = Max ⎜
⎜ × ∗ ⎟ (7)
fj − fj ⎟
n −
parameters, with the intersection of each well and chemical given ⎝ ⎠
Wj
as Vij , one can evaluate the rank of each well via TOPSIS by the
following formula in maximizing the Cj∗ (Hwang and Yoon, 1981): j=1


n 2 3. Results and discussion
(V
j=1 ij
− Vj− )
Cj∗ =   (5)
n n 2 According to WQI results, 20 wells out of 92 wells are classified
(V
j=1 ij
− Vj∗ )2 + (V − Vj− )
j=1 ij in the Excellent water classification with WQI values smaller than
50, while 45 wells are obtained in the range of 50 < WQI < 100 which
where Vj∗ , Vj− , and Cj∗ are weighted ideal and negative-ideal solu-
are considered as good water. Based on WQI indexes and wells’
tion, and similarity ratio, respectively.
locations in the case study, most of the wells with WQI < 100 are
Another prioritization method is CP which evaluates the wells’
located in central and eastern part of the Varamin plain. In addition,
ranks by Formula (6). Among the wells, the best one has the least
the number of wells which are classified in poor water and very
distance from the ideal point (Zeleny, 1973).
poor water are six and five wells, respectively. In total, 16 wells out
⎡ ⎛ ⎞p ⎤1/p of 92 wells of the case study are determined unsuitable for drinking
purpose (Figs. 2 and 3).
⎢ n ⎜ ⎟ ⎥
⎢ ⎜ Wj ∗
fj − fij⎟ ⎥ The ranks of wells in each MCDM and WQI method for four
Lp (Welli ) = ⎢


⎜ n × ∗ ⎟ ⎥ (6)
fj − fj ⎟ ⎥
− classes of water quality are shown by Fig. 4. In WQI method, rank
⎣ j=1 ⎝  ⎠ ⎦ for each well in each class is defined according to its WQI value. For
Wj
instance, the well with the lowest value of WQI in excellent class
j=1
is ranked as the first one, while the well with the highest WQI is
where Lp (Welli ) represents the distance of the well to the ideal solu- ranked 51 (among 51 wells) in the same class. This method of rank-
tion. fj∗ and fj− bare ideal and negative-ideal solutions, respectively. ing is repeated for other classes. In good water range, the well in
S. Zahedi et al. / Ecological Indicators 78 (2017) 243–252 249

Fig. 5. Coefficient of determination for classes (a) Excellent water (0–75), (b) Good water (50–150), (c) Poor Water (100–250), (d) very poor water (200–350).

the first rank has the lowest WQI and the 49th well has the highest CP was worse than well number 66, well number 35, well number
WQI value. 83, and well number 61 which were classified in good water class
The results of unfit water for drinking purpose can be found with WQI values of 50.61, 50.80, 50.92 and 51.29, respectively. This
in Fig. 1 of the Supplementary material and are not interfered in well’s rank in TOPSIS was 8. The factor which increases the rank of
the rest of calculation due to the fact that this class is not used for this well in CP model is its higher amount of HCO3 (220 mg/L) in
drinking purpose. comparison to other wells. It should be noticed that HCO3 has the
According to Fig. 4(a) in the TOPSIS and the CP for different val- least effect on WQI values because of its low relative weight (0.027).
ues of p, wells’ ranks faced considerable fluctuations. For instance, This well has the highest level of HCO3 among the 25 top-ranked
although WQI value for the well number 51 was equal to 42.96, wells of excellent water class yet, after performing normalization of
which gave the 8th rank to the well, and placed it in the excellent the dataset for TOPSIS method, the mentioned considerable change
water class, its rank descended to 15, 15, 26 and 12 for CP(p = 1), occurred. In addition, the high level of Ca and Cl in this well is
CP(p = 2), CP(p = ∞) and TOPSIS, respectively. The changes in rank- another reason for its rank degradation. According to high amount
ing lists were more touchable for wells which belonged to the of SO4 in wells number 66, 35, 83 and 61 and the significance of
excellent and good water classes and the wells in classical WQI this parameter in WQI method with high weight of 0.138, the good
range. In another example, although the well number 71 with WQI rank of well number 71 is acceptable. The results of comparison
value of 46.69 was classified in the excellent water class, its rank in between these four wells can be seen in Table 4.
250 S. Zahedi et al. / Ecological Indicators 78 (2017) 243–252

Table 4
The results of comparison between available four wells of new method range.

WELLS K Na Mg Ca HCO3 Cl SO4 pH T.D.S NO3 WQI Value

num71 2.74 64.4 24.3 50.7 220 89.7 69.7 7.7 383 0.1 46.7
num66 3.13 59.1 25.8 48.1 148 69.1 129 8.25 490 0.3 50.6
num35 3.13 107 15.2 25.9 124 70.9 155 8.27 521 0.1 50.8
num83 3.13 69.2 20.9 46.3 174 67.4 112 7.83 548 0.4 50.9
num61 2.74 98.9 22.6 32.5 146 46.1 162 8.55 498 0.01 51.3

Table 5
Some examples of changes in ranks of the wells according to the classes.

WQI Range WELL’s Code WQI Compromise programing (CP) TOPSIS Average

num(␹) Value Rank p=1 Rank p=2 Rank p=∞ Rank C* Rank Rank

0–75 num47 42.3 7 3.14 14 1.14 14 0.72 23 0.76 17 13


num87 45.6 13 2.75 6 1.01 6 0.6 13 0.85 10 7
num51 43 8 3.24 15 1.15 15 0.75 26 0.83 12 14
num58 44.1 11 3.09 12 1.05 9 0.59 12 0.9 7 8
num71 46.7 16 4.01 25 1.44 28 0.84 33 0.87 8 22
num85 51.4 26 2.91 8 1.07 11 0.56 9 0.57 27 15
num64 49.8 20 3.62 21 1.42 27 0.85 34 0.78 15 25
num66 50.6 21 3.86 24 1.23 18 0.47 5 0.64 23 16

50–150 num31 51.3 5 1.79 4 0.76 6 0.54 10 0.99 8 6


num85 51.4 6 1.06 1 0.41 1 0.24 1 0.99 5 1
num55 98.5 45 5.75 48 1.95 48 0.92 40 0.39 46 48
num17 100 47 4.79 43 1.66 40 0.81 33 0.61 40 42
num23 100 46 5.31 47 1.95 47 1 44 0.54 41 47
num27 109 48 4.74 42 1.71 43 0.96 42 0.23 48 46

100–250 num23 100 1 2.5 3 1.42 4 1 4 0.99 1 3


num17 100 2 1.64 2 0.82 2 0.7 2 0.98 2 1
num27 109 3 1.41 1 0.57 1 0.36 1 0.78 3 1
num5 227 7 8.53 8 2.74 8 1 4 0.03 8 8
num8 236 8 7.54 7 2.53 7 1 4 0.04 7 7

200–350 num5 227 1 3.83 3 1.69 3 1 2 0.99 1 2


num8 236 2 2.35 1 1.1 1 0.83 1 0.98 2 1
num21 279 3 2.95 2 1.37 2 1 2 0.78 3 3

300 ≤ WQI num11 381 6 3.91 11 1.6 11 1 12 0.99 5 8


num19 391 8 1.6 1 0.58 1 0.31 1 0.94 8 6
num30 522 11 4.79 12 1.7 13 1 12 0.6 14 13
num16 536 13 3.81 10 1.31 10 0.7 7 0.61 13 11

Some examples of changes in ranks of the wells according to the class related to unfit water for drinking purpose is not used in this
classes are also presented by Table 5. part.
According to Fig. 4(b), in comparison to well number 17, According to the results of regression coefficient, in each class of
although the well number 55 has lower amount of WQI (better rank water quality there were more than two MCDM methods with high
in WQI method), it gains worse rank in other four MCDMs which regression coefficient in comparison to WQI method. In addition,
stems from lack of equilibrium between its water quality param- the regression coefficient values of CP method with p = 1 and 2 were
eters. The amount of NO3 in well number 55 is 0.9 mg/L which is greater than 0.9. Thus capability of these two methods for ranking
three times larger than NO3 in well number 17. In WQI method, the wells in this case study should be taken into account. Furthermore,
NO3 concentration after deviation to the WHO standards’ value will the regression coefficient in TOPSIS method decreased by raising
decrease to 45 mg/L. Due to its low amount, effectiveness of this the number of wells in each class. TOPSIS in classes of 100–250 and
parameter in WQI evaluation will be reduced. However, MCDMs 200–350 has higher regression coefficient than WQI method. CP
increased the effectiveness of NO3 because throughout the nor- method for p = ∞ has the lowest regression coefficient among other
malization step, the normalized amount of NO3 in well number 3 methods. Hence, increase of “p” parameter was not appropriate for
became 3.5 times larger than the well number 17. Thus, the results reliable prioritizing of wells’ water quality.
of MCDMs seem more realistic. Such explanations revealed that Li et al. (2013a,c) reported efficiency of MCDMs in sensitiv-
although some wells were placed in the lower ranks based on the ity analysis of criteria weights. The current research showed that
original WQI ranges, they might even be placed in the higher ranks MCDMs are also capable of validating WQI classes and investigating
of the table after the abovementioned modifications and utilizing sensitivity of the MCDMs to number of alternatives. In compar-
the MCDMs. There is no remarkable change in Fig. 4(c) and (d) but ison to similar studies, application of WQI in some studies has
some changes exist in different used methods. faced different issues. According to the results of Sadat-Noori et al.
Due to the differences between five methods (WQI, TOPSIS and (2013), although some wells belongs to different classes, yet the
CP for p = 1, 2 and ∞) in prioritizing wells in the case study, the interpolation results based on the Kriging method put them in a
averaging method is used for comparing MCDMs vs. WQI method. same category of WQI class. On the other hand, by employment
The regression coefficients between averaging method and the five of MCDMs, there would be no same contradiction or paradoxical
above-mentioned methods in four classes (excellent water, good outcome in classification of wells. The results of the present study
water, poor water and very poor water) are shown via Fig. 5. The are in line with the above-mentioned concepts. Jamshidzadeh and
S. Zahedi et al. / Ecological Indicators 78 (2017) 243–252 251

Mirbagheri (2010) just analyzed wells separately with correlation Backman, B., Bodis, D., Lahermo, P., Rapant, S., Tarvainen, T., 1998. Application of a
coefficient between different parameters and concluded that the groundwater contamination index in Finland and Slovakia. Environ. Geol. 36,
55–64, http://dx.doi.org/10.1007/s002540050320.
separation assessment would not be practical for appraising water Bozorg-Haddad, O., Azarnivand, A., Hosseini-Moghari, S.M., Loáiciga, H.A., 2016a.
quality in a case study. However, in the current study not only role Development of a comparative multiple criteria framework for ranking pareto
of each well was considered in the class of groundwater quality, optimal solutions of a multiobjective reservoir operation problem. J. Irrig.
Drain. Eng. 142 (7), 04016019.
but the wells are also ranked based on multiple criteria analysis of Bozorg-Haddad, O., Azarnivand, A., Hosseini-Moghari, S.M., Loáiciga, H.A., 2016b.
harmful chemical parameters for drinking purpose. WASPAS application and evolutionary algorithm benchmarking in optimal
reservoir optimization problems. J. Water Resour. Plann. Manage., 04016070.
Brown, R.M., McClelland, N.I., Deininger, R.A., Tozer, R.G., 1970. A water quality
4. Conclusion index: do we dare? Water Sewage Works 117, 339–343.
Chang, C.L., Lin, Y.T., 2014. A water quality monitoring network design using fuzzy
The main objective of the current study was to explore a solution theory and multiple criteria analysis. Environ. Monit. Assess. 186 (10),
6459–6469.
to mitigate probable errors of WQI method in classification of water Chitsaz, N., Azarnivand, A., 2016. Water scarcity management in arid regions based
quality classes. The most striking features to emerge from the study on an extended multiple criteria technique. Water Resour. Manage., 1–18.
are as follows: Chitsaz, N., Banihabib, M.E., 2015. Comparison of different multi criteria
decision-making models in prioritizing flood management alternatives. Water
Resour. Manage. 29 (8), 2503–2525.
(1) In WQI method, some impreciseness may occur like a decrease Chitsaz, N., Malekian, A., 2016. Development of a risk-based multi-criteria
in WQI just because of low quantity of a parameter with high approach for watershed prioritization with consideration of soil erosion
alleviation (case study of Iran). Environ. Earth Sci. 75 (22), 1448.
relative weight or less computational effectiveness of a param- Dahiya, S., Singh, B., Gaur, S., Garg, V.K., Kushwaha, H.S., 2007. Analysis of
eter with low relative weight but high quantity. Furthermore, groundwater quality using fuzzy synthetic evaluation. J. Hazard. Mater. 147
small changes in a parameter may cause an overwhelming shift (3), 938–946.
ESRI, 1999. ARCVIEW GIS v.3.2, Environmental Systems Research. Institute Inc.
in WQI results.
Goovaerts, P., 1997. Geostatistics for Natural Resources Evaluation. Oxford
(2) Introducing MCDMs instead of WQI was the achievement of the University Press, New York, pp. p. 483.
proposed research. MCDMs can reanalyze groundwater quality Hernandez, E.A., Uddameri, V., 2010. Selecting agricultural best management
practices for water conservation and quality improvements using Atanassov’s
classes of the case study after modifying boundary values of
intuitionistic fuzzy sets. Water Resour. Manage. 24 (15), 4589–4612.
WQI classes. Hosseini-Moghari, S.M., Ebrahimi, K., Azarnivand, A., 2015. Groundwater quality
(3) These results indicated that by employment of such MCDMs as assessment with respect to fuzzy water quality index (FWQI): an application of
TOPSIS and CP which use different assumptions as similarity to expert systems in environmental monitoring. Environ. Earth Sci. 74 (10),
7229–7238, http://dx.doi.org/10.1007/s12665-015-4703-1.
the ideal solution and compromise solution, some of the wells Hosseinifar, S.J., Mirzaei Aminiyan, M., 2015. Hydrochemical characterization of
could gain higher scores despite existing in the lower quality groundwater quality for drinking and agricultural purposes: a case study in
class of general WQI range. rafsanjan plain, Iran. Water Qual. Expo. Health 7531, http://dx.doi.org/10.
1007/s12403-015-0169-3.
(4) The results also demonstrate that TOPSIS and CP yield differ- Howard, K.W.F., Howard, K.K., 2016. The new Silk Road Economic Belt as a threat to
ent ranking list for each class. TOPSIS for the classes with low the sustainable management of Central Asia’s transboundary water resources.
number of wells performed better in comparison to the classes Environ. Earth Sci. 75, 976, http://dx.doi.org/10.1007/s12665-016-5752-9.
Hwang, C.L., Yoon, K., 1981. Multiple attribute decision making: Methods and
with high quantity of wells. Moreover, increases in value of the applications, a state of the art survey. Springer, New York.
sensitivity parameter of CP technique reduced the accuracy of Jamshidzadeh, Z., Mirbagheri, S.A., 2010. Evaluation of groundwater quantity and
results. quality in the Kashan Basin, Central Iran. Desalination 270 (2011), 23–230.
Khan, A., Maity, K., 2016. Application of MCDM-based TOPSIS method for the
optimization of multi quality characteristics of modern manufacturing
Owing to proper application of MCDMs, it can be a practical idea processes. Int. J. Eng. Res. Afr. 23, 33–51.
to use and develop MCDM-based water quality assessment models Krishan, G., Singh, S., Kumar, C.P., Gurjar, S., Ghosh, N.C., 2016. Assessment of water
quality index (WQI) of groundwater in Rajkot District, Gujarat, India. J. Earth
for future researches. Sci. Clim. Change 7, 341, http://dx.doi.org/10.4172/2157-7617.1000341.
Lermontov, A., Yokoyama, L., Lermontov, M., Machado, M.A.S., 2009. River quality
Acknowledgments analysis using fuzzy water quality index: Ribeira do Iguape river watershed,
Brazil. Ecol. Indic. 9 (6), 1188–1197.
Li, P., Qian, H., Wu, J., 2010. Groundwater quality assessment based on improved
The authors gratefully acknowledge the helpful comments of water quality index in Pengyang County, Ningxia, Northwest China. E-J. Chem.
Gordana J. Devic on an earlier draft of this manuscript. Suggestions 7, S209–S216, http://dx.doi.org/10.1155/2010/451304.
Li, P., Qian, H., Wu, J., 2011. Hydrochemical formation mechanisms and quality
from two anonymous reviewers and the respectable Associated assessment of groundwater with improved TOPSIS method in Pengyang
Editor are also greatly appreciated. We also acknowledge Dr. A.M. County Northwest China. E-J. Chem. 8 (3), 1164–1173, http://dx.doi.org/10.
Zahedi for his helps in editing English writing of the paper. 1155/2011/251918.
Li, P., Wu, J., Qian, H., 2012. Groundwater quality assessment based on rough sets
attribute reduction and TOPSIS method in a Semi-arid Area, China. Environ.
Appendix A. Supplementary data Monit. Assess. 184 (8), 4841–4854, http://dx.doi.org/10.1007/s10661-011-
2306-1.
Li, P., Qian, H., Wu, J., Chen, J., 2013a. Sensitivity analysis of TOPSIS method in
Supplementary data associated with this article can be found, water quality assessment: I. Sensitivity to the parameter weights. Environ.
in the online version, at http://dx.doi.org/10.1016/j.ecolind.2017. Monit. Assess. 185, 2453–2461, http://dx.doi.org/10.1007/s10661-012-2723-9.
03.015. Li, P., Wu, J., Qian, H., 2013b. Assessment of groundwater quality for irrigation
purposes and identification of hydrogeochemical evolution mechanisms in
pengyang county, China. Environ. Earth Sci. 69 (7), 2211–2225, http://dx.doi.
References org/10.1007/s12665-012-2049-5.
Li, P., Wu, J., Qian, H., Chen, J., 2013c. Sensitivity analysis of TOPSIS method in
Aghazadeh, N., Asghari-Mogaddam, A., 2010. Assessment of groundwater quality water quality assessment II: sensitivity to the index input data. Environ. Monit.
and its suitability for drinking and agricultural uses in the oshnavieh area, Assess. 185, 2463–2474, http://dx.doi.org/10.1007/s10661-012-2724-8.
northwest of Iran. J. Environ. Prot. 1, 30–40, http://dx.doi.org/10.4236/jep. Li, P., Wu, J., Qian, H., Lyu, X., Liu, H., 2014. Origin and assessment of groundwater
2010.11005. pollution and associated health risk: a case study in an industrial park,
Annapoorna, H., Janardhana, M.R., 2015. Assessment of groundwater quality for northwest China. Environ. Geochem. Health 36, 693–712, http://dx.doi.org/10.
drinking purpose in rural areas surrounding a defunct copper mine. Aquat. 1007/s10653-013-9590-3.
Procedia 4, 685–692. Li, P., Qian, H., Howard, K.W.F., Wu, J., 2015. Building a new and sustainable Silk
Azarnivand, A., Banihabib, M.E., 2016. A multi-level strategic group decision Road economic belt. Environ. Earth Sci. 74 (10), 7267–7270, http://dx.doi.org/
making for understanding and analysis of sustainable watershed planning in 10.1007/s12665-015-4739-2.
response to environmental perplexities. Group Decis. Negot., 1–20, http://dx.
doi.org/10.1007/s10726-016-9484-8.
252 S. Zahedi et al. / Ecological Indicators 78 (2017) 243–252

Li, P., Li, X., Meng, X., Li, M., Zhang, Y., 2016a. Appraising groundwater quality and Rizwan, R., Gurdeep, S., 2010. Assessment of ground water quality status by using
health risks from contamination in a semiarid region of Northwest China. Water Quality Index method in Orissa, India. World Appl. Sci. J. 9 (12),
Expo.Health 8 (3), 361–379, http://dx.doi.org/10.1007/s12403-016-0205-y. 1392–1397.
Li, P., Wu, J., Qian, H., 2016b. Hydrochemical appraisal of groundwater quality for Sadat-Noori, S.M., Ibrahimi, K., Liaghat, A.M., 2013. Groundwater quality
drinking and irrigation purposes and the major influencing factors: a case assessment using the water quality index and GIS in Saveh-Nobaran aquifer,
study in and around Hua County, China. Arabian J. Geosci. 9 (1), 15. Iran. Environ. Earth Sci. 71 (9), 3827–3843.
Madani, K., Lund, J.R., 2011. A Monte-Carlo game theoretic approach for Saeedi, M., Abessi, O., Sharifi, F., Meraji, H., 2010. Development of groundwater
multi-criteria decision making under uncertainty. Adv. Water Resour. 34 (5), quality index. Environ. Monit. Assess. 163, 327–335.
607–616. Srinivas, R., Bhakar, P., Pratap Singh, A., 2015. Groundwater Quality Assessment in
Milovanovic, M., 2007. Water quality assessment and determination of pollution some selected area of Rajasthan, India Using Fuzzy Multi-Criteria Decision
sources along the Axios/Vardar River, Southeastern Europe. Desalination 213, Making Tool. Aquat. Procedia 4, 1023–1030, International Conference on
159–173. Water Resources, Coastal and Ocean Engineering (ICWRCOE 2015).
Mosaferi, M., Mahmoudi, M., Pourakbar, M., Sheykholeslami, S., 2015. Vasanthavigar, M., Srinivasamoorthy, K., Vijayaragavan, K., Rajiv-Ganthi, R.,
Groundwater quality assessment in a volcanic region iran. Proceedings of the Chidambaram, S., Anandhan, P., Manivannan, R., Vasudevan, S., 2010.
14th International Conference on Environmental Science and Technology. Application of water quality index for groundwater quality assessment:
Narany, T.S., Ramli, M.F., Aris, A.Z., Sulaiman, W.N.A., Fakharian, K., 2015. Thirumanimuttar sub-basin, Tamilnadu, India. Environ. Monit. Assess. 171
Groundwater irrigation quality mapping using geostatistical techniques in (1–4), 595–609, http://dx.doi.org/10.1007/s10661-009-1302-1.
Amol-Babol Plain, Iran. Arabian J. Geosci. 8 (2), 961–976, Rhodes, Greece, 3–5 WHO, 2004. Guidelines for Drinking Water Quality: Training Pack. WHO, Geneva,
September 2015. Switzerland.
Nikolić, D., Milošević, N., Mihajlović, I., Živković, Ž., Tasić, V., Kovačević, R., Petrović, Wu, J., Sun, Z., 2016. Evaluation of shallow groundwater contamination and
N., 2010. Multi-criteria analysis of air pollution with SO2 and PM10 in urban associated human health risk in an alluvial plain impacted by agricultural and
area around the copper smelter in Bor, Serbia. Water Air Soil Pollut. 206 (1–4), industrial activities, Mid-west China. Expo. Health 8 (3), 311–329, http://dx.
369–383. doi.org/10.1007/s12403-015-0170-x.
Ning, S.K., Chang, N.B., 2005. Screening the relocation strategies of water quality Zeleny, M., 1973. Compromise programming. In: Cochrane, J.L., Zeleny, M. (Eds.),
monitoring stations by compromise programming. JAWRA J. Am. Water Multiple Criteria Decision Making. University of South Carolina Press,
Resour. Assoc. 41 (5), 1039–1052. Columbia, South Carolina.
Pomerol, J.C., Barba-Romero, S., 2000. Multicriterion Decision in Management:
Principles and Practice. Kluwer Academic, New York.

Potrebbero piacerti anche