Assessment of Pulsating Organic and
Nutrient Load in a River System, Using Multivariate Statistical Techniques: A
Case Study of the
Mahanadi River in Odisha, India
Pradyumna Dash1* 
, Pramode Kumar Prusty2, Swoyam
Prakash Rout3
1Department of Chemistry, Christ College, Cuttack, 753008, Odisha India
2Department of Forest and Environment, Government of Odisha,
Bhubaneswar India
3Department of Chemistry, Utkal University, Vani Vihar,
Bhubaneswar, 751004, Odisha India
*Corresponding Author E-mail: pradyumnadash007@gmail.com
Natural processes
and the anthropogenic activities are the major source of surface water
pollution. In recent years various statistical methods have been used to
formulate environmental classification in river water systems. Multivariate
statistical techniques, such as cluster analysis (CA), and principal component
analysis (PCA) / factor analysis (FA) were applied for the assessment of
organic and nutrient load in the Mahanadi river, Odisha (India). Agglomerative
hierarchical cluster analysis (AHC) grouped sampling sites into three clusters,
i.e., relatively less polluted (LP), moderately polluted (MP) and highly
polluted (HP) sites, based on the similarity of water quality characteristics.
Factor analysis/principal component analysis, applied to the data sets of the
three different groups obtained from cluster analysis, resulted in three, two
and one principal component (PC) respectively having Eigen value >1. The PCs
obtained from factor analysis indicates that the increase in load of nutrients
is mainly attributed to agriculture runoff, industrial effluents and regional
anthropogenic activities. The magnitude of BOD with respect to total N and P
demonstrates the intensity of organic pollution in the riverine water. Biplot
between principal components establish the mode of association of parameters
and their interrelationships for evaluating the water quality of the river
system. This study represent the necessity and usefulness of multivariate
statistical techniques for assessment of large and complex databases to extract
better information about the water quality of surface water bodies, the design
of sampling and analytical protocols and the effective measures to control /
management of pollution load in the surface water.
KEYWORDS: Mahanadi river, Industrial /Urban sewage, Cluster analysis,
Principal Component analysis / Factor analysis, Water quality.
1. INTRODUCTION:
The degradation of surface water
quality is a major concern in various developing countries including India.
Since rivers are the main inland water bodies that often carry large municipal
sewage, industrial waste water and agricultural runoffs, are generally enriched
in nutrients as compared to other environments. The deterioration of water
quality leading to water resource problems are mainly attributed to
anthropogenic influences from both point and non-point source pollution as well
as change in natural process such as precipitation inputs and erosion (Carpenter
et al ; 1998 and Jarvi et al ;1998).
.
The quality of a river at any
point reflects several major influences including the lithology of the basin,
atmospheric inputs, climatic conditions and anthropogenic inputs (Bricker
and Jones, 1995). On the other hand, rivers play a major role in
assimilation or transporting municipal and industrial wastewater and runoff
from agricultural land (Sundaray et al;2006). Municipal and industrial
wastewater discharge constitutes a constant polluting source, whereas surface
runoff is a seasonal phenomenon, largely affected by the climatic condition
within the basin .However, due to spatial and temporal variations in water
quality, a regular monitoring program that can provides a representative and
reliable estimation of the quality of surface water is highly essential. (Dixon
and Chiswell, 1996).
The application of different
multivariate statistical techniques, such as cluster analysis (CA) and
principal component analysis (PCA)/ factor analysis (FA) helps in deriving
hidden information from the data set and the possible influences of environment
on water quality (Spanos et al ; 2003). FA/ PCA attempts to explain the
correlation between the observations in terms of underlying factors, which are
not directly observable. The first stage in factor analysis is to generate a
correlation matrix for all the variables. The correlation matrix thus obtained
is used to determine the Eigen values and factor loadings. The factor loading
are used to measure the correlation between the variables and factors. Thus
these statistical methods can be effectively used to interpret the complex data
matrices for better understanding of the water quality and ecological status of
the studied systems. This also allows the identification of possible factors/
sources that influence water systems and offers a valuable tool for reliable
management of water resources as well as faster solution to pollution problems
(Vega et al., 1998; Adams et al., 2001; Reghunath et al., 2002; Simeonova et
al., 2003; Simeonov et al., 2004).The Mahanadi watershed is the most
developed and urbanised region in the state of Odisha. Rapid urbanisation and
industrialization in its neighbourhood in the recent past have greatly
influenced the surface water quality of the watershed. The uncontrolled and
improper disposal of solid and toxic waste from municipal, industrial,
agricultural and other human activities are increasingly deteriorating the
water quality of the riverine system. The enhancing water pollution not only
causes deterioration of water quality but also threatens human health and
balance of aquatic ecosystem, economic development and social prosperity.
In the present study, mean value
of a large data matrix, obtained during 4-year (2007 to 2010) monitoring
program is subjected to different multivariate statistical techniques to
extract information about the similarities or dissimilarities between sampling
sites, identification of water quality variables responsible for spatial and
temporal variations in nutrient and organic loadings in the Mahanadi river
basin. The specific objectives are to
i) Identify
several zones with different water quality.
ii) Extract
the parameters that are responsible for nutrient and organic loadings in the
river system.
iii) Generate
a good approach to assess the water quality reasonably that can be helpful to
the managers to take the valuable measures to manage the water resource
effectively.
2. STUDY SITES:
The Mahanadi river system is the
third largest in the peninsula of India and the largest river in Orissa state.
The basin extends over an area approximately 141,600 km2 out of
which 65,628 km2 lies in Odisha occupying 42.15% of the state
geographical area. Out of its total length of 851 km it covers 491 km in the
state of Odisha with a peak discharge of 44,740 m3/s (Konhauser
et al.1997). The geographical co-ordinates of Mahanadi basin (figure-1)
lying in Odisha lies between 82010' to 86050' East
longitudes and 19030' to 22015' North latitudes. River
Mahanadi originates from a small pool located at about 6 km from Pharsiya
village in the Amarkantak hills of Bastar Plateau, which lies on the extreme
south east of Raipur district of Chhattisgarh state. Ib, Ong, Tel are the main
tributaries while Kathajodi , Kuakhai, Devi, and Birupa are the major
distributaries of Mahanadi in Odisha. It flows over different geological
formations of Eastern Ghats and adjacent areas and joins the Bay of Bengal at
Paradip after dividing into different branches in the deltaic area. Along it’s
course the Mahanadi receives untreated or partially treated effluents from
various large and small scale industries as listed in Table-1 and urban centres
such as Sambalpur, Jagatpur and Paradip in Odisha (Radhakrishna 2001).
Figure-1-Map of Mahanadi basin
in Odisha (India), indicating monitoring stations
Figure-2-Schematic Diagram of
the Mahanadi river system indicating major towns and monitoring stations
It also receives a large amount
of agricultural run-off along its course. Human influences are pronounced at
Sambalpur, Cuttack and Paradip, where the proliferation of industries and sewer
discharges is prominent. Although a lot of work has been carried out on the
role of different urban and industrial effluents upon the water quality of the
Mahanadi river system (Upadhayay 1988; Chakrapani and Subramanian 1990,
1993; Das et al.2002; Nanda and Tiwari 2001; Nayak et al. 2001; Dixit et
al. 2013), the study only deals with the factors responsible for organic
and nutrient loads in the river system using multivariate statistical
techniques. The present study deals with the evaluation of 10 different
physico-chemical parameters from 15 monitoring stations from the Hirakud
reservoir (MR1) to Paradip D/s (MR15), covering a short stretch of the river
course in the state of Odisha. The physico-chemical parameters such as NO2-N,
NO3-N, NH4-N, NH3-N, TKN and total phosphorus
(TP) were analyzed, along with pH, DO, BOD, and COD in order to assess the
nutrient and organic load in the riverine system. The details of the monitoring
stations and major urban settlement along the Mahanadi river are given in the
schematic diagram (figure-2)
3. MATERIAL AND METHODS:
3.1 Sampling and Parameters:
Water samples were collected
from 15 stations along the course of the Mahanadi river (MR) system, starting
from the Hirakud reservoir (MR1) to Paradip D/s (MR15). The sampling strategy
was designed in such a way to cover a wide range of determinants at key sites
that accurately represent the water quality of the river systems and account
for tributary inputs that can have important impacts upon downstream water
quality. Only certain limited parameters accounting organic and nutrient loads
from 15 monitoring stations were analysed in four different seasons from 2007
to 2010. The mean value of the data sets was taken into consideration for
evaluating the pollution load in the water system. The measured parameters
include field pH, dissolved oxygen (DO), biochemical oxygen demand (BOD),
chemical oxygen demand (COD), ammonical (NH4-N), free ammonia (NH3-N),
nitrate (NO3−N), nitrite (NO2−N), total
Kjeldahl nitrogen (TKN), total phosphorous (TP). The summary basic statistics
of the dataset is presented in (Table-2).Statistical analyses, viz. cluster
analyses (CA), principal component analysis (PCA) / factor analysis (FA), were
carried out for 10 parameters at 15 monitoring stations and the data were
processed using XLSTAT-11.5 statistical package.
Table -1-Major industries operating in
Mahanadi Basin in Odisha.
|
Name of the industries and Location
|
Products
|
|
A) Mahanadi
|
|
|
Arati Steel Ltd., Athagarh.
|
Steel
|
|
Hindalco Industries Ltd., Sambalpur.
|
Power
|
|
Hindalco Industries Ltd., (Smelter)
Sambalpur.
|
Aluminium Smelter Plant
|
|
Orissa Power generation Cooperation
(OPGC), Banharpali.
|
Power
|
|
Bargarh Co-operative Sugar Mills Ltd.,
Bargarh.
|
Sugar
|
|
ACC Cement, Bargarh.
|
Cement
|
|
Bijayananda Co-operative Sugar Mills
Ltd., Bolangir.
|
Sugar
|
|
Nayagarh Sugar Complex Ltd., Nayagarh.
|
Sugar
|
|
Cosboard Industries Ltd., Cuttack.
|
Cardbpard and paper
|
|
SMV Beverages Pvt Ltd.,Cuttack.
|
Softdrinks
|
|
Paradip Phosphate Ltd., Jgatsinghpur.
|
Phosphatic Fertilisers
|
|
IFFCO Ltd., Jagatsinghpur.
|
Phosphatic Fertilisers
|
|
Skol Breweries Ltd., Jgatsinghpur
|
Beer
|
|
B) Ib
|
|
|
Vedanta Aluminium Ltd., Jharsuguda
|
Alminium Smelter Plant
|
|
Sterlite Energy (P) Ltd., Jharsuguda
|
Power
|
|
TRL Krosaki Refractories., Belpahar
|
Refractories
|
|
Ultar Tech Cement Ltd., Jharsuguda
|
Cement
|
|
Bhusan Power and Steel Ltd., Jharsuguda
|
Iron and Steel, Power
|
|
C) Birupa
|
|
|
Indian Metal and Ferro Alloys Ltd. (CPP),
Choudwar
|
Power
|
|
Indian Metal and Ferro Alloys Ltd.
(Ferroalloys), Choudwar
|
Chargechrome
|
3.2 Statistical analysis:
For a better understanding of
the natural and anthropogenic fluxes responsible for the characterization of
water quality in the Mahanadi river system, factor analysis (FA) /principal
component analysis (PCA) and cluster analyses (CA) were carried out from the
mean value of data set obtained from four different years from 2007 to 2010.
The factor analyses (FA)/ principal component analysis (PCA) were carried out
with data set having Eigen values >1 that are considered to have significant
influence towards the geo-chemical processes (Sahu et al. 1998; Panigrahy et
al. 1999). The agglomerative hierarchical clustering was carried out from
data normalized to a zero mean and using Euclidian distances as a measure of
similarity (Massart and Kaufman 1983). Ward’s method was selected
because it possesses a small space-distorting effect and accesses more
information on cluster content (Angelidis and Aloupi 2000). The results
indicate that all the applied statistical techniques are useful in offering
reliable classification of surface water in the whole region and make it
possible to design a future spatial sampling strategy in an optimal method,
which can reduce the number of sampling sites.
3.2.1 Cluster Analysis:
Cluster analysis (CA) is used to
develop meaningful aggregations or groups of entities based on a large number
of interdependent variables. The resulting clusters of objects should exhibit
high internal (within - cluster) homogeneity and high external (between
clusters) heterogeneity. Agglomerative hierarchical clustering (AHC)
analysis is the most common approach, which provides intuitive similarity
relationships between any one sample and the entire data set, and is typically
illustrated by a dendrogram (tree diagram) (McKenna, 2003). The
dendrogram provides a visual summary of the clustering processes, presenting a
picture of the groups and their proximity with a dramatic reduction in
dimensionality of the original data. The Euclidian distance usually gives the
similarity between two samples and a distance can be represented by the
difference between analytical values from the samples (Otto, 1998). In
this study, AHC analysis was performed on the normalized data set by means of
the Ward’s method, using Euclidean distances as a measure of similarity.
The Ward’s method uses an analysis of variance approach to evaluate the
distances between clusters in an attempt to minimize the sum of squares (SS) of
any two clusters that can be formed at each step. The spatial variability of
water environment quality in the whole river basin was determined from CA,
which divides a large number of objects into smaller number of homogenous
groups on the basis of their internal correlations.
3.2.2 Principal component analysis / factor
analysis:
Factor analysis, which includes
PCA is a very powerful technique applied to reduce the dimensionality of a
dataset consisting of a large number of interrelated variables, while retaining
as much as possible the variability presented in dataset. This reduction is
achieved by transforming the dataset into a new set of variables namely the
principal components (PCs), which are orthogonal (non-correlated) and are
arranged in decreasing order of importance. Mathematically, the PCs are
computed from covariance or other cross-product matrix, which describes the
dispersion of the multiple measured parameters to obtain Eigen values and Eigenvectors.
PCA also attempts to explain the correlations between the observations in terms
of the underlying factors, which are not directly observable (Yu et al. 2003).
This study comprises application of multivariate statistical techniques to
analyze water quality dataset obtained from the Mahandi River in Odisha.
Table-2.Minimum, Maximum, Mean and Standard
deviation of water quality parameters at different sampling points from 2007 to
2010
|
Variable
|
Observations
|
Minimum
|
Maximum
|
Mean
|
Std. deviation
|
|
pH
|
15
|
7.420
|
7.900
|
7.734
|
0.108
|
|
DO
|
15
|
7.170
|
8.200
|
7.723
|
0.290
|
|
BOD
|
15
|
0.950
|
2.600
|
1.436
|
0.472
|
|
COD
|
15
|
7.400
|
40.320
|
13.767
|
8.139
|
|
NH4-N
|
15
|
0.590
|
1.370
|
0.927
|
0.224
|
|
NH3-N
|
15
|
0.026
|
0.065
|
0.043
|
0.010
|
|
NO3-N
|
15
|
1.251
|
2.812
|
1.970
|
0.434
|
|
NO2-N
|
15
|
0.282
|
0.625
|
0.444
|
0.094
|
|
TKN
|
15
|
3.630
|
15.910
|
9.215
|
3.545
|
|
TP
|
15
|
0.014
|
0.390
|
0.080
|
0.093
|
4. RESULTS AND DISCUSSION:
Spatial similarities and site
grouping:
In this study, sampling sites
classification was performed by the use of cluster analysis. The relationships
among the stations were obtained through cluster analyses using Ward’s method
(linkage between groups), with Euclidian distance as a similarity measure and
were synthesized into dendogram plots (figure-3). Since we used hierarchical
agglomerative cluster analysis, the number of clusters was also decided by
water environment quality, which is mainly effected by land use and industrial
activities. The physico-chemical parameters like pH, DO, BOD, COD, NH4-
N, NH3-N, NO3-N, NO2-N, total Kjeldahl –N
(TKN) and total phosphorus (TP) were used as variables and showed a sequence in
their association, displaying the information as degree of contamination.
Grouped stations under each cluster can be seen in (figure-3). Basing upon the
result of the cluster analysis, conclusions were drawn.
Agglomerative Hierchichal
Cluster Analysis (AHC)
Figure-3Dendrogram showing
clustering of sampling sites according to surface water quality characteristics
of the Mahanadi river basin.
Cluster-
I (MR 1-2-3-6-7-8-9-10-14):
This cluster mainly includes
sites namely Hirakud resorvior (MR1), Power channel D/s (MR2), Sambalpur U/s
(MR3), Huma (MR6), Sonepur U/s (MR7), Sonepur D/s (MR8), Tikarpara
(MR9), Narasinghpur (MR10) and Paradip U/s (MR14) . The impact of human
activities on the riverine ecosystem in this region is relatively low. Although
the agricultural runoff and the direct discharge of untreated domestic
wastewater in certain sampling sites contaminated the water, cluster I
corresponds to less Polluted (LP) site, because the inclusion of sampling
location suggests that the self purification and assimilative capacity of the
river are quite strong.
Cluster- II (MR 5-11-12-13):
This cluster mainly include
sites such as Sankarmath (MR-5) further downstream to Sambalpur, Cuttack U/s
(MR11), Cuttack D/s (MR12) and Gatiroutpatna (MR13) further downstream to Cuttack
are classified as Moderately Polluted (MP). All these sampling sites are
located near major urban settlement or waste water outfall. The degradation in
water quality mainly attributed to influx of untreated or partially treated
municipal sewage from the Sambalpur and Cuttack townships. The nutrient
discharged through the effluents are partially diluted due to the maximum input
of fresh water into the riverine system leading to moderate nutrient and
organic loadings at these sites. The distribution of nitrite among these sites
suggests that terrestrial inputs are very low and localized effects appear to
be more pronounced.
Cluster- III ( MR 4-15):
This cluster mainly includes
Sambalpur D/S (MR4) and Paradip D/s (MR15). These sites are classified as
highly polluted (HP). These stations receive pollution mostly from domestic
wastewater, wastewater treatment plants and industrial effluents located in
city. Both the sites are responsible for high concentration of organic load
into the river which is reflected by low concentration of DO. The decrease in
DO level at MR15 is due to the presence of fertiliser industry at its upstream
and at MR4 is due to untreated domestic and municipal waste water which
directly discharge acidic effluents into the riverine system.
The results indicate that the CA
technique is useful in offering reliable classification of surface waters in
the whole region and will make it possible to design a future spatial sampling
strategy in an optimal manner, which can reduce the number of sampling stations
and associated costs. There are other reports ( Kim et al.,2005) where
similar approach has successfully been applied to water quality programme.
Table-3Summary statistics after Principal
component analysis of LP site
|
Variable
|
Observations
|
Minimum
|
Maximum
|
Mean
|
Std. deviation
|
|
pH
|
9
|
7.420
|
7.820
|
7.717
|
0.122
|
|
DO
|
9
|
7.420
|
8.200
|
7.869
|
0.211
|
|
BOD
|
9
|
0.950
|
1.700
|
1.209
|
0.249
|
|
COD
|
9
|
7.400
|
15.200
|
10.498
|
2.626
|
|
NH4-N
|
9
|
0.590
|
1.020
|
0.801
|
0.151
|
|
NH3-N
|
9
|
0.026
|
0.053
|
0.038
|
0.009
|
|
NO3-N
|
9
|
1.725
|
2.391
|
2.025
|
0.227
|
|
NO2-N
|
9
|
0.380
|
0.540
|
0.457
|
0.052
|
|
TKN
|
9
|
4.800
|
15.910
|
10.047
|
3.451
|
|
TP
|
9
|
0.014
|
0.093
|
0.050
|
0.027
|
Figure-4 Scree plot for LP site.
4.1 Data structure determination
and source identification
Principal component
analysis/factor analysis was performed on the normalized data sets separately
for the three different regions, viz., LP, MP and HP as delineated by CA
techniques to compare the composition structure between analyzed water samples
and the factors loadings of each variable. PCA of the three data sets yield
four PCs for LP and MP sites while two PCs were obtained for HP with each cases
having Eigen values >1. An Eigen value gives a measure of the significance for
the factor, which with highest Eigen value is the most significant. Eigen
values of 1.0 or greater are considered significant (Kim and Mueller 1978)
and the factor loadings were classified as ‘strong’, ‘moderate’ and ‘weak’,
corresponding to absolute loading values of >0.75, 0.75–0.50 and 0.50–0.30
respectively (Liu et al. 2003).
The summary statistics of the
data set (Table-3) pertaining to LP sites, generates eight PCs (Table-4) out of
which three PCs are significant having Eigen values >1 as shown in the scree
plot of LP sites (figure-4) and having cumulative variance of 80.43%. Among the
three significant principal components, PC1, with Eigen value 3.590 and
explaining 35.90 % of the total variance, has moderate positive loadings on
BOD, COD, NH3-N, TP and strong positive loading pH, NO3-N,
and NO2-N. This factor represents the contribution of non-point
source pollution from agricultural lands. In these areas, farmers uses the
nitrogenous fertilizer, which undergo nitrification processes and the rivers
receive nitrate nitrogen via groundwater leaching. The strong positive
correlation between pH with NO2-N (r =0.58) and NO3-N(r
=0.52) (Table- 5), suggests that the increase in pH value of the water is
associated with increasing nitrogenous species. In PC1, COD-BOD correlation
reflects that the correlation coefficient and correlation were significant at
those locations where river generally receives organic waste (BOD). PC2, with
Eigen value 2.562 and explaining 25.62% of the total variance has moderate positive
loadings on COD, moderate negative loading on DO and strong positive loading on
TKN with significant negative correlation of TKN with DO. This may be
attributed to increasing organic nitrogen load that results in depletion of DO
in these sites. PC3 with Eigen value 1.891 and explaining 18.90% of the total
variance, has strong positive loadings on NH4-N and moderate
positive loading on DO and NH3-N with no significant correlation.
Both PC2, and PC3 represent organic pollution from domestic waste and nonpoint
source pollution.
The biplot between PC1 and PC2,
where the scores of samples drawn and the loadings of variables have been
plotted (figure-5) reveals similar water quality statistics between MR3 and
MR13 having increasing load of NH4-N in the first quadrant. The
second quadrant containing only MR6 (Huma), show close association between NH3-N,
pH, BOD, COD and TKN. This shows both organic (BOD) and total Kjeldahl nitrogen
(TKN) loading at this site and may be attributed to direct release of untreated
or partially treated industrial effluents and municipal waste into water
system. The third quadrant containing MR1, MR8 and MR9 show close association
of NO3-N, NO2-N, and DO. This can be interpreted as the
utilization of inorganic nutrients supporting the photosynthetic process that
helps to regenerate DO. The forth quadrant showing stations MR2, MR7 and MR10
exhibit almost similar water quality statistics and remain almost unpolluted
throughout the year. This shows that the self purification and assimilative
capacity of the river in these sites are quite strong.
Table-4-Factor Loading / Principal
Component analysis of LP sites
|
Variables
|
PC1
|
PC2
|
PC3
|
PC4
|
PC5
|
PC6
|
PC7
|
PC8
|
|
pH
|
0.792
|
0.259
|
-0.311
|
-0.449
|
-0.040
|
0.016
|
0.070
|
-0.011
|
|
DO
|
0.291
|
-0.622
|
0.560
|
0.294
|
0.262
|
0.185
|
0.162
|
-0.014
|
|
BOD
|
0.728
|
0.335
|
0.089
|
0.045
|
0.567
|
-0.050
|
-0.125
|
-0.090
|
|
COD
|
0.543
|
0.653
|
-0.248
|
0.362
|
-0.081
|
0.257
|
-0.080
|
0.085
|
|
NH4-N
|
-0.148
|
0.275
|
0.857
|
0.121
|
-0.368
|
0.016
|
-0.115
|
-0.073
|
|
NH3-N
|
0.623
|
0.184
|
0.716
|
-0.100
|
0.064
|
-0.179
|
0.013
|
0.139
|
|
NO3-N
|
0.760
|
-0.589
|
0.029
|
-0.153
|
-0.217
|
0.042
|
-0.047
|
-0.019
|
|
NO2-N
|
0.793
|
-0.535
|
-0.019
|
-0.163
|
-0.230
|
0.051
|
-0.046
|
-0.022
|
|
TKN
|
0.061
|
0.890
|
0.285
|
-0.283
|
-0.092
|
0.087
|
0.151
|
-0.057
|
|
TP
|
0.658
|
0.184
|
-0.288
|
0.583
|
-0.235
|
-0.203
|
0.106
|
-0.046
|
|
Eigenvalue
|
3.590
|
2.562
|
1.891
|
0.916
|
0.701
|
0.188
|
0.105
|
0.046
|
|
Variability (%)
|
35.903
|
25.623
|
18.907
|
9.161
|
7.013
|
1.883
|
1.046
|
0.465
|
|
Cumulative %
|
35.903
|
61.525
|
80.433
|
89.594
|
96.606
|
98.489
|
99.535
|
100.000
|
Table-5-Correlation matrix between
physico-chemical parameters of LP sites
|
Variables
|
pH
|
DO
|
BOD
|
COD
|
NH4-N
|
NH3-N
|
NO3-N
|
NO2-N
|
TKN
|
TP
|
|
pH
|
1
|
-0.232
|
0.584
|
0.514
|
-0.359
|
0.358
|
0.515
|
0.576
|
0.333
|
0.410
|
|
DO
|
-0.232
|
1
|
0.186
|
-0.269
|
0.190
|
0.422
|
0.502
|
0.447
|
-0.443
|
0.005
|
|
BOD
|
0.584
|
0.186
|
1
|
0.552
|
-0.123
|
0.606
|
0.233
|
0.264
|
0.285
|
0.409
|
|
COD
|
0.514
|
-0.269
|
0.552
|
1
|
-0.033
|
0.204
|
-0.004
|
0.060
|
0.455
|
0.715
|
|
NH4-N
|
-0.359
|
0.190
|
-0.123
|
-0.033
|
1
|
0.521
|
-0.181
|
-0.208
|
0.468
|
-0.149
|
|
NH3-N
|
0.358
|
0.422
|
0.606
|
0.204
|
0.521
|
1
|
0.377
|
0.371
|
0.406
|
0.195
|
|
NO3-N
|
0.515
|
0.502
|
0.233
|
-0.004
|
-0.181
|
0.377
|
1
|
0.997
|
-0.409
|
0.333
|
|
NO2-N
|
0.576
|
0.447
|
0.264
|
0.060
|
-0.208
|
0.371
|
0.997
|
1
|
-0.367
|
0.374
|
|
TKN
|
0.333
|
-0.443
|
0.285
|
0.455
|
0.468
|
0.406
|
-0.409
|
-0.367
|
1
|
-0.021
|
|
TP
|
0.410
|
0.005
|
0.409
|
0.715
|
-0.149
|
0.195
|
0.333
|
0.374
|
-0.021
|
1
|
Figure-5 Biplot of PC1 Vs PC2 of
LP sites
The summary statistics of the
data set (Table-6) pertaining to MP sites, generates three principal
components out of which two PCs are significant having Eigen value >1 as
shown in the scree plot of MP sites (figure-6) and having cumulative variance
of 90.78%. Among the two significant PCs, (Table-7), PC1, with Eigen value
5.054 and explaining 50.54% of the total variance, has strong positive
loadings on BOD, COD, NO3 - N, NO2-N, moderate positive
loading on pH, NH3-N, TKN and strong negative loading of DO. This
factors represents the inclusion of both organic and nutrient load into the
riverine system from municipal, domestic waste and non-point source pollution
from agricultural land containing nitrogenous fertilizer. The strong positive
correlation between pH with TKN (r = 0.98) and NH3-N (r = 0.74)
(Table-8), suggests that the high value pH of the water is associated with
increasing load of organic nitrogen. In PC1 both COD and BOD shows strong
positive correlation with NO3-N and NO2-N, which reflects
that the nutrient plays a major role in increasing the organic load of the MP
sites. PC2, with Eigen value 4.024 and explaining 40.24% of the total variance
has moderate positive loadings on NH3-N, TKN and strong positive
loading on pH and NH4-N. This can be established as strong positive
correlation between pH with TKN (r = 0.98), NH3-N (r = 0.74) and NH4-N
(r = 0.69). This may be attributed to increasing load of nitrogenous species
and organic nitrogen load from both point and non-point source pollution that
results in increase of pH level of water in these sites.
The biplot between PC1 and PC2
(figure-7), reveals close association between DO and NH4-N at MR11
in the first quadrant. The second quadrant containing sites MR13 and MR12 shows
a close association between NH3-N, pH, COD and TKN. This shows how
pH of these sites changes with increase in organic nitrogen load in the river
system. The third quadrant does not include any sampling site but the data set
represent close association of NO3-N, NO2-N, and DO, which
may be attributed to the utilization of inorganic nutrients supporting the
photosynthetic process that helps regenerate DO. The forth quadrant showing
only site MR5 where nutrient load is only due to the increasing concentration
of TP.
Table-6 Summary Statistics after principal
component analysis of MP sites.
|
Variable
|
Observations
|
Minimum
|
Maximum
|
Mean
|
Std. deviation
|
|
pH
|
4
|
7.670
|
7.900
|
7.760
|
0.112
|
|
DO
|
4
|
7.570
|
7.870
|
7.660
|
0.141
|
|
BOD
|
4
|
1.120
|
2.150
|
1.610
|
0.463
|
|
COD
|
4
|
8.900
|
17.370
|
13.063
|
3.502
|
|
NH4-N
|
4
|
0.820
|
1.370
|
1.110
|
0.231
|
|
NH3-N
|
4
|
0.038
|
0.053
|
0.046
|
0.006
|
|
NO3-N
|
4
|
1.251
|
1.793
|
1.484
|
0.239
|
|
NO2-N
|
4
|
0.282
|
0.405
|
0.341
|
0.057
|
|
TKN
|
4
|
6.122
|
9.126
|
7.503
|
1.535
|
|
TP
|
4
|
0.034
|
0.390
|
0.133
|
0.172
|
Table- 7 Factor Loading/Principal Component
analysis of MP sites
|
Variables
|
PC1
|
PC2
|
PC3
|
|
pH
|
0.546
|
0.755
|
-0.362
|
|
DO
|
-0.786
|
0.236
|
0.571
|
|
BOD
|
0.870
|
-0.469
|
0.151
|
|
COD
|
0.996
|
0.082
|
0.033
|
|
NH4-N
|
-0.152
|
0.986
|
-0.074
|
|
NH3-N
|
0.611
|
0.705
|
0.359
|
|
NO3-N
|
0.892
|
-0.369
|
0.260
|
|
NO2-N
|
0.834
|
-0.529
|
0.157
|
|
TKN
|
0.694
|
0.681
|
-0.233
|
|
TP
|
-0.127
|
-0.907
|
-0.401
|
|
Eigenvalue
|
5.054
|
4.024
|
0.922
|
|
Variability (%)
|
50.538
|
40.238
|
9.223
|
|
Cumulative %
|
50.538
|
90.777
|
100.000
|
Figure-6 Scree plot for MP
sites
Figure-7 Biplot of PC1 Vs PC2
of MP sites
Lastly, the summary statistics
of the data sets (Table-9) pertaining to HP sites, generates one PC having
Eigen value >1 i.e 10.00 as shown in the scree plot of HP sites (figure-8)
and explaining 100% of total variance. It shows strong positive loadings on pH,
BOD, COD, NH3-N, NO3-N, NO2-N, TKN, TP and
strong negative loading on DO (Table-10). The principal component suggest high
concentration of nutrient load and greater intensity of organic pollution at
this site which can be interpreted as untreated wastewater and sewage disposal
from both Sambalpur and Paradip Cities and the influx of nitrogenous effluents
from peripheral fertiliser industries. The strong negative loading on DO, can
also be attributed to anaerobic conditions in river from the loading of high
dissolved organic matter. Through the PCA, the sources of the pollutants were
identified in the three zones. As mentioned above, it can be helpful to the
government and managers, who can lay down different regulations and policies in
three zones respectively.
Table-8 Correlation matrix between
physico-chemical parameters of MP sites
|
Variables
|
pH
|
DO
|
BOD
|
COD
|
NH4-N
|
NH3-N
|
NO3-N
|
NO2-N
|
TKN
|
TP
|
|
pH
|
1
|
-0.458
|
0.067
|
0.595
|
0.688
|
0.737
|
0.115
|
-0.001
|
0.978
|
-0.609
|
|
DO
|
-0.458
|
1
|
-0.709
|
-0.745
|
0.310
|
-0.110
|
-0.640
|
-0.691
|
-0.518
|
-0.343
|
|
BOD
|
0.067
|
-0.709
|
1
|
0.833
|
-0.606
|
0.255
|
0.989
|
0.998
|
0.250
|
0.254
|
|
COD
|
0.595
|
-0.745
|
0.833
|
1
|
-0.073
|
0.679
|
0.867
|
0.792
|
0.740
|
-0.214
|
|
NH4-N
|
0.688
|
0.310
|
-0.606
|
-0.073
|
1
|
0.576
|
-0.518
|
-0.660
|
0.583
|
-0.845
|
|
NH3-N
|
0.737
|
-0.110
|
0.255
|
0.679
|
0.576
|
1
|
0.379
|
0.193
|
0.821
|
-0.861
|
|
NO3-N
|
0.115
|
-0.640
|
0.989
|
0.867
|
-0.518
|
0.379
|
1
|
0.980
|
0.308
|
0.117
|
|
NO2-N
|
-0.001
|
-0.691
|
0.998
|
0.792
|
-0.660
|
0.193
|
0.980
|
1
|
0.182
|
0.311
|
|
TKN
|
0.978
|
-0.518
|
0.250
|
0.740
|
0.583
|
0.821
|
0.308
|
0.182
|
1
|
-0.612
|
|
TP
|
-0.609
|
-0.343
|
0.254
|
-0.214
|
-0.845
|
-0.861
|
0.117
|
0.311
|
-0.612
|
1
|
Table-9-Summary Statistics after principal
component analysis of HP sites.
|
Variable
|
Observations
|
Minimum
|
Maximum
|
Mean
|
Std. deviation
|
|
pH
|
2
|
7.750
|
7.770
|
7.760
|
0.014
|
|
DO
|
2
|
7.170
|
7.220
|
7.195
|
0.035
|
|
BOD
|
2
|
1.620
|
2.60
|
2.110
|
0.692
|
|
COD
|
2
|
19.450
|
40.320
|
29.885
|
14.757
|
|
NH4-N
|
2
|
1.060
|
1.190
|
1.125
|
0.091
|
|
NH3-N
|
2
|
0.046
|
0.065
|
0.055
|
0.013
|
|
NO3-N
|
2
|
2.578
|
2.812
|
2.695
|
0.165
|
|
NO2-N
|
2
|
0.567
|
0.625
|
0.596
|
0.041
|
|
TKN
|
2
|
3.630
|
14.165
|
8.897
|
7.449
|
|
TP
|
2
|
0.060
|
0.154
|
0.106
|
0.066
|
Figure-8 Scree plot for HP sites
Factor Loading/Principal:
Component analysis of HP site:
Table-10:
|
Variables
|
PC1
|
|
pH
|
1.000
|
|
DO
|
-1.000
|
|
BOD
|
1.000
|
|
COD
|
1.000
|
|
NH4-N
|
1.000
|
|
NH3-N
|
1.000
|
|
NO3-N
|
1.000
|
|
NO2-N
|
1.000
|
|
TKN
|
1.000
|
|
TP
|
1.000
|
|
Eigenvalue
|
10.000
|
|
Variability (%)
|
100.000
|
|
Cumulative %
|
100.000
|
5. CONCLUSIONS:
In this present study,
multivariate statistical techniques i.e CA and PCA / FA were used to evaluate
the organic and nutrient loadings of the Mahanadi river basin. Hierarchical
cluster analysis grouped 15 sampling sites into three distinct clusters on the basis
of similar water quality characteristics. Based on the result obtained, it is
possible to design an optimal sampling strategy, which could reduce the number
of sampling stations and associate costs. Also this analysis segregate LP sites
into four different zones, MP sites into three distinct zones and HP sites into
one distinct zone with different water quality. Although the factor/principle
component analysis did not result in a significant data reduction, it helped
extract and identify the factors/sources responsible for variations in river
water quality at three zones. The major pollutants responsible for organic load
in all the three zones are contributed by local anthropogenic activities rather
than agricultural/ land drainage. The intensity of microbial activities and the
influx of organic sewage are reflected through the high BOD, NH3-N,
NO3-N, NO2-N, TKN values for cluster-III in HP, which are
more than the permissible limit for drinking water. The inverse relationship
between DO with BOD and DO with total nitrogen and total phosphate in HP
implies that the organic nitrogen plays a major role in the depletion of DO in
the river systems. The factors obtained from principal component analysis
indicate that the parameters responsible for water quality variations are
mainly related to organic pollution (point source: domestic wastewater) in
relatively less polluted (LP) sites, organic pollution (point source: domestic
wastewater) and nutrients (non-point sources: agriculture and orchard
plantations) in moderately polluted (MP) sites, and organic pollution and
nutrients (point sources: domestic wastewater, wastewater treatment plants and
industries) in highly polluted (HP) sites in the basin. With serious situation
of water pollution in the Mahanadi watershed, the management of water quality
of the different zones is becoming more and more important as well as the
planning of the whole watershed. Thus, this study illustrates the usefulness of
multivariate statistical techniques for analysis and interpretation of complex
data sets in water quality assessment, identification of pollution
sources/factors and understanding spatial variations in water quality for
effective management of river water quality.
6. ACKNOWLEDGEMENTS:
We are thankful to SPCB, Odisha,
for providing relevant water quality data. We also thank the Head, Department
of Chemistry, Utkal University, for his assistance, and the staff members of
the Department for their cooperation and valuable suggestion. The first author
is grateful to Mr. Alok. Mishra, Head of the Department of Chemistry Christ
College, Cuttack for providing necessary laboratory facilities while carrying
out this study.
7. REFERENCES:
[1] Angelidis MO, Aloupi M.
Geo-chemical study of coastal sediments influencing river-transported pollution
Southern Evoikos Golf, Greece. Mar Pollut Bull. 40(1); 2000: 77–82.
[2] Adams S, Titus R, Pietesen K.,
Tredoux G. and Harris C. Hydrochemical characteristic of aquifers near
Sutherland in the Western Karoo, South Africa. Journal of Hydrology. 241; 2001:
91-103.
[3] Bricker OP. and Jones BF. Main
factors affecting the composition of natural waters. In: Salbu B., Steinnes E.
(Eds.), Trace Elements in Natural Waters. CRC Press, Boca Raton.1995;
FL: pp. 1-5.
[4] Carpenter SR, Caraco N E,
Correll DL, Howarth RW and Smith VH. Nonpoint pollution of surface waters with
phosphorus and nitrogen. Ecological Applications. 8(3) ;1998: 559–568.
[5] Chakrapani GJ and Subramanian V.
Preliminary studies on the geo-chemistry of the Mahanadi basin,India. Chem
Geol. 81; 1990: 241–253.
[6] Chakrapani GJ and Subramanian
V. Rates of erosion and sedimentation in the Mahanadi river basin. India. J
Hydrol. 149:1993: 39–48.
[7] Das J, Sahoo RK. and Sinha BK..
Urban ground water pollution: a case study in Cuttack city, India. Ground Water
Monitor Remed. 22(3); 2002: 95–103.
[8] Dixit PR, Kar B, Chhatopadhyay
P and Panda CR. Seasonal variation of the physicochemical properties of water
sample in Mahanadi estuary, East coast of India, Journal of Environ. Protection.
4; 2013: 443-448.
[9] Dixon W and Chiswell B. Review
of aquatic monitoring program design. Water Research 30; . 1996:
1935-1948.
[10] Jarvie, HP, Whitton, BA and
Neal C. Nitrogen and phosphorus in east coast British rivers: speciation,
sources and biological significance. Science of the Total Environment. 210/211;
1998: 79–109.
[11] Kim J O And Mueller C W.
Introduction to factor analysis: what it is and how to do it. Quantitative
applications in the social sciences.Saga publication.New bury park.1978.
[12] Kim JH, Kim RH, Lee J, Cheong
TJ, Yum BW and Chang HW. Multivariate statistical analysis to identify the
major factors governing groundwater quality in the coastal area of Kimje, South
Korea. Hydrological Processes. 19; 2005: 1261-1276.
[13] Konhauser KO, Powell MA, Fife W,
Longstaffle FJ, and Tripathy S. Trace element geochemistry of river sediment,
Orissa state, India. J Hydrol, 193; 1997: 258–269.
[14] Liu CW, Lin KH, and Kuo YM.
Application of factor analysis in the assessment of groundwater quality in a
Blackfoot disease area in Taiwan. Science of the Total Environment. 313; 2003:
77–89.
[15] Massart DL and Kaufman L. The
Interpretation of Analytical Chemical Data by the Use of Cluster Analysis
Wiley. NewYork. 1983.
[16] McKenna Jr J.E. An enhanced
cluster analysis program with bootstrap significance testing for ecological
community analysis. Environmental Modelling and Software. 18 (3); 2003: 205-220
[17] Nanda SN and Tiwari TN. Effect
of discharge of municipal sewage on the quality of River Mahanadi at Sambalpur.
Indian J Environ Protect. 21(4); 2001: 336–343.
[18] Nayak BB, Das J, Panda UC and
Acharya BC. Industrial effluents and municipal sewage contamination of Mahanadi
estuarine water, Orissa. Proceedings Published Allied Publishers. New Delhi.
2001.pp. 77–86.
[19] Otto M.. Multivariate methods.
In: Kellner R, Mermet JM, Otto M, Widmer HM. (Eds.), Analytical Chemistry.
Wileye VCH, Weinheim. 1998.
[20] Panigrahy PK, Das J, Das SN and
Sahoo RK. Evaluation of the influence of various physico-chemical parameters on
coastal water quality, around Orissa, by factor analysis. Indian J Mar Sci. 28;
1999: 360–364.
[21] Radhakrishna I. Saline fresh
water interface structure in Mahanadi delta region, Orissa, India. Environ Geol
40(3); 2001: 369–380.
[22] Reghunath R, Murthy TRS and
Raghavan BR. The utility of multivariate statistical techniques in
hydrogeochemical studies: an example from Karnataka, India. Water Research. 36;
2002: 2437-2442.
[23] Sahu KC, Panda UC and Mohapatra
DM. Geo-chemistry and mineralogy of sediments in Rushikulya estuary, East coast
of India. Chem Environ Res. 7(1–2); 1998: 77–92.
[24] Simeonov V, Simeonova P and
Tsitouridou R. Chemometric quality assessment of surface waters: two case
studies. Chemical and Engineering Ecology. 11 (6); 2004: 449-469.
[25] Simeonov V, Stratis JA, Samara
C, Zachariadis G, Voutsa D, Anthemidis A, Sofoniou M, and Kouimtzis Th.
Assessment of the surface water quality in northern Greece. Water Research.
37(17); 2003: 4119–4124.
[26] Spanos T, Simenov V, Stratis
J, and Xristina X. Assessment of water quality for human consumption .Michrochim
Acta. 141; 2003: 35-40.
[27] Sundaray SK., Panda UC, Nayak
BB. and Bhatta D. Multivariate statistical techniques for the evaluation of
spatial and temporal variations in water quality of the Mahanadi
river–estuarine system (India) – a case study. J. Environ. Geochem. Health. 28;
2006: 317–330.
[28] Upadhyay S. Physico-chemical
characteristics of the Mahanadi estuarine eco-system, east coast of India. Indian
J Mar Sci. 17; 1988: 19–23.
[29] Vega M, Pardo R, Barrado E
and Deban L. Assessment of seasonal and polluting effects on the quality of
river water by exploratory data analysis. Water Research. 32; 1998: 3581-3592.
[30] Yu SX., Shang JC, Zhao JS, and
Guo HC. Factor analysis and dynamics of water quality of the Songhua River
Northeast China. Water, Air and Soil Pollution. 144(1–4); 2003: 159–169.