Professional Documents
Culture Documents
Mustapha (2013) Surface Water Quality
Mustapha (2013) Surface Water Quality
Mustapha (2013) Surface Water Quality
DOI 10.1007/s12517-012-0731-2
ORIGINAL PAPER
Received: 6 June 2012 / Accepted: 29 October 2012 / Published online: 25 November 2012
# Saudi Society for Geosciences 2012
lithology of the basin, atmospheric inputs, climatic conditions, In recent years, there has been an increasing amount of
and anthropogenic inputs (Gamble and Babbar-Sebens 2012). literature on the use of principal component analysis and
Since the sources of river run-off are a combination of surface factor analysis (PCA/FA) in pollution sources apportion-
run-off, rainfall, groundwater, etc., and a river contains sea- ment. The application of different multivariate statistical
sonal variability features, the water quality of a river is sus- tools such as PCA/FA helps in the understanding of surface
ceptible to pollution and thus, it is essential to construct water pollution sources, allowing the identification of pos-
trustworthy information on water pollution sources and con- sible factors that influence water chemistry and offer valu-
trolling the adverse effects. able tools for the management of water resources (Simeonov
There are various ways through which pollutants enter et al. 2003; Singh et al. 2004; Shrestha and Kazama 2007;
into the surface water: point and non-point sources. The Juahir et al. 2009; Hanh et al. 2010; Palma et al. 2010; Aris
point sources includes waste water effluent from municipal et al. 2012a, b; Cieszynska et al. 2012; Chabukdhara and
and industrial run-off, and leachate from waste disposal site, Nema 2012). Several researchers used correlation matrix,
run-off from animal feed lots, run-off from mines, oil fields, principal component, and/or factor analysis to identify the
un-sewerage industrial site, storm sewer outfalls from cities, relationship of variables and water quality source’s appor-
overflow of combined storm and sanitary sewers, run-off tionment. Kumar and Reddy (2008) studied the relationship
from construction sites, among others. Non-point pollution of physiochemical parameters in Buckingham canal, India;
is a diffused source of pollution, having no definite or the study revealed a significant relationship exists between
specific location where discharge into a body of water. For nutrient and organic related variables. Shrestha and Kazama
example, farm field, parking lots, construction sites, and (2007) Huang et al. (2010) and Juahir et al. (2009) studied
agricultural field. These sources are often difficult to mon- spatial variability of surface water quality and sources
itor, regulate or treat. The non-point source of water pollu- apportionment and classified the studied water bodies into
tion are often intermittent and linked to seasonal agricultural high pollution site (HP), moderate pollution site (MP), and
activities, they are derived from extensive area of land and low pollution site (LP). Factor analysis revealed that the
are transported overland, underground, or through the atmo- pollution levels in the three zones were mainly influenced
sphere to receiving water (Carpenter et al. 1998). These by natural sources (temperature and river discharge) and
sources of pollutants may accumulate large quantity of anthropogenic source’s (industrial, municipal, and agricul-
nutrients enrichment and stimulates eutrophication process tural run-off). Onojake et al. (2011) in their studies, they
and bioaccumulation of organic and inorganic compounds. discovered that Rivers in Delta state of Nigeria were heavily
Further, it may result in alteration of tropic interaction polluted as a result of industrial discharge and municipal
among both aquatic flora and fauna (Mustapha and Aris waste (anthropogenic source of pollution). They used PCA/
2012a). The sources and pathways of urban and agricultural FA to identify the latent factors that explain the chemistry of
pollutants are given in Table 1. the surface water, PCA/FA yielded three PC’s with more
than 82 % variance. Vidal et al. (2000) used PCA/FA for the above mean sea level, covering an area of about 150 km2
study of water resources contamination due to the use of (Mustapha et al. 2012a). The basin is one of the most signif-
livestock slurries as fertilizer, PCA/FA reveled the sampled icant in supporting residential, industrial, and agricultural
data was related to dilution of water with strong saline activities in Kano metropolitan with a population of more than
content and redox condition. Equally, PCA/FA was used 500,000 inhabitants. The climate of the basin is strongly
on the chemical data to distinguish the contribution of influenced by the tropical continental and maritime air masses.
natural and anthropogenic sources of ground water pollution The two air masses control the climate making it dry and wet
in Grombalia Plain in North eastern Tunisia. Three indepen- type (Mustapha et al. 2012b). The pre-Cambrian rock of the
dent factors were extracted and explained 77.86 % of the basement complex which comprises of gneisses, amphibo-
total variance which reveled in the presence of two groups lites, marble, and older granite underlies larger part of Nigeria,
of waters: mineralization and artificial recharge water type including 80 % of the Jakara Basin. The climate of the region
(Tlili-Zrelli et al. 2012). Brahim et al. 2011 applied PCA/FA controls the amount of water that is available both on the
with varimax rotation on nine normalized physiochemical surface and at the sub-surface at any given time within a water
variables. The varimax factors obtained showed that year. Two major system of crop production have been seen to
groundwater variation is controlled by water-rock interac- be practiced within the region. These are rain-fed agriculture
tion processes and anthropogenic influence through domes- and irrigation farming. Farming is done in the city core areas,
tic water discharge. Moreover, PCA/FA results obtained wedge areas, and corridors out of the city and on the periphery.
showed that water quality variations in the semi-arid region Urban and peri-urban irrigation provide the populace with
of India are mainly controlled by the hydro-geochemical vegetables and fruits within the urban environment for house-
processes (weathering and dissolution) and anthropogenic hold consumption as well as for sell to the rapidly growing
processes (domestic and agricultural activities). The study population. Different range of crops and fruits are produced,
concluded that carbonate, silicate weathering, halite disso- including maize, beans, tomatoes, pumpkins, melon, pepper,
lution, and ion exchanges are the dominant processes occur- cabbage, lettuce, onion, and carrots.
ring in the study area (Singh et al. 2012). The chemistry of
the ground water from the phreatic aquifer of Ouargla,
Algeria was studied by Semar et al. 2012. PCA/FA with Materials and methods
varimax rotation was applied on the datasets with the aim of
identifying the relationship between physiochemical param- Sampling and analytical procedure
eters and the source of groundwater quality variations. PCA/
FA provides a cumulative variance explained by 67 % with A total of 405 water samples were collected from various
axes controlled by sodium chloride (NaCl), calcium sulfate locations along the main river and tributaries. Ten, eight,
(CaSO4), bicarbonate (HCO3−), and nitrite (NO2−). Koklu et and nine sampling points were strategically located along
al. (2010) focused on water quality of Melen River, Turkey; Jakara, Tsakama, and Gama-Kwari Rivers, respectively
they used PCA/FA and identified factors responsible for the (Fig. 1). The location of the sampling points was recorded
variation in water quality. They concluded that all stations using Garmin global positioning system and water samples
under study were affected by agricultural and industrial were collected every other day from 1st December, 2010 to
pollution due to the intensive agricultural activities through 31st December, 2010.
application of fertilizer. Liu et al. (2009) used PCA/FA to Sampling points were selected to represent different land
detect the latent pollution sources of Songhua River and use activities around the study area. Prior to the sampling
identified five latent factors: organic, inorganic, petrochem- procedure, the bottles were washed with the distilled water
icals, physiochemical, and heavy metals pollution sources; and rinsed three times with the river water three times to
they concluded that Songhua River water quality was main- avoid possible contamination. The water samples were col-
ly controlled by domestic wastewater and industrial effluent. lected below 15 cm of the subsurface layer to avoid the
The present study evaluates the physicochemical and colloidal layer as this may influence the concentration of
bacteriological parameters of Jakara watershed using PCA/ certain parameters. Water samples were collected in tripli-
FA and Pearson’s product–moment correlation to identify cates to estimates the variability resulting from the sampling
the pollution status and probable sources of pollution. and analytical procedures using 200 mL polyethylene bot-
tles and stored in an ice box before transporting to the
laboratory. The sample was preserved in a refrigerator at
Study area description 4 ° C to avert microbial growth, flocculation, and reduce any
adsorption on the polythene bottles. Samples were preserved
Jakara Basin lies on longitude 8° 45 E and 12° 05E, latitude according to standard methods for the examination of water
10° 30 N and 13° 02 N. It is located at about 481 m (1,580 ft) and waste water (APHA 2005) for subsequent analysis.
4906 Arab J Geosci (2013) 6:4903–4915
Fig. 1 Study area and sampling 8.454725 8.534440 8.614155 8.693870 8.773585
point’s location
12.179478
12.179478
N
12.123304
12.123304
12.067130
12.067130
12.010956
12.010956
Legend
11.954782
11.954782
0 1.5 3 6 9 12 15
Kilometers
8.454725 8.534440 8.614155 8.693870 8.773585
Thirteen physiochemical and biological water quality Pearson product moment correlation analysis
parameters were selected for analysis: this being dissolved
oxygen (DO), 5-day biochemical oxygen demand (BOD5), A bivariate correlation analysis produces Pearson’s correlation
chemical oxygen demand (COD), suspended solids (SS), pH, coefficient (r). The correlation coefficient gives a mathemat-
ammonia-nitrogen (NH3NL), dissolved solids (DS), total sol- ical value for measuring the strength and magnitude of the
ids (TS), nitrates (NO3), chloride (Cl), phosphates (PO4), linear relationship between two bivariates. The value of rela-
Escherichia coli (E. coli) and fecal coliform bacteria (FCB). tionship takes values ranging from −1 to +1, where +1 repre-
Water temperature, pH, and DO of the river water sam- senting an absolute perfect positive linear relationship (as x
ples were measured in-situ, using thermo-probe (Model increase, y increase). Zero representing no linear relationship
TL1-A), a pre-calibrated microprocessor pH meter (Hanna between the bivariates (x and y have no pattern), whereas −1
Instruments, model: pH211) and DO meter (4500-OG) re- represents an absolute inverse relationship between bivariates
spectively. SS, TS, and DS were separated and measured by (as x increases, y decreases). The sign in front of the correla-
gravimetric: filtering the water through a 0.45 μm filter tion coefficient determines the direction of the relationship. A
paper and determine according to a standard procedure plus sign denotes positive relationship and a minus sign
(APHA 2005). Cl was determined by argentometric titration denotes negative correlation. The Pearson correlation coeffi-
methods. NH3NL was measured by chemical titration using cient (r) can be calculated using the equation below:
salicylate test kits method. NO3 and PO4 were determined P P P
n xy ð xÞð yÞ
using a molecular absorption spectrophotometer. E. coli and r ¼ rhffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi ð1Þ
FCB counts were based on multiple tube fermentation tech- P 2 P 2 ih P 2 P 2i
n x ð x Þ n y ð yÞ
niques (APHA 2005). BOD5 was measured after 5 days of
incubation at 20 °C in tightly stoppered BOD bottles in the Where r is the correlation coefficient value, x and y are
dark (APHA 2005). COD measurements were performed by the bivariate.
the potassium dichromate method (K2CrO7).
In order to provide greater data confidence from the
analytical procedure regarding bias and variability, appro- Principal component analysis and factor analysis
priate quality control and quality assurance (QA/QC) on
water samples were ensured. The QA/QC was followed to PCA/FA are statistical approaches that can be used to ana-
ensure that data products are of documented high quality lyze interrelationships among a large number of variables
and reproducible. The overall data quality is assessed and to explain these variables in terms of their common
through precision, accuracy, and comparability (Aris et al. underlying dimension by providing empirical estimates of
2012a). The analytical precision was assessed by the use of the structure of variables (Singh et al. 2004). The objective
control chart and blank samples. The same laboratory equip- of PCA/FA is to reduce the number of variables necessary to
ments were used for all samples collected in order to control describe the observed variation within a datasets. PCA/FA
variability from sampling irregularities. lessens the number of variables understudy and discovers
Arab J Geosci (2013) 6:4903–4915 4907
the structure between the variables (Pejman et al. 2009). distributions of mean (Gaur and Gaur 2006). The hypothesis
This is achieved by forming linear combinations of the can be tested as follows:
variables (components) that describe the distribution of the
Ho : μ1 þ μ2 þ μ3 ::::::μk ð3Þ
data (Mustapha and Aris 2012b). The linear combinations
are derived from some measures of association such as
correlation or covariance matrix. PCA converts the original Ho Not all means are equal
dataset which comprises of measurements for each variable The F statistics calculate the ratio between the variance
measured for each sample and converts them to an equal due to difference between group and error variance. The
number of composite variables (Mustapha and Nabegu larger the F ratio, the greater is the difference between groups
2011; Mustapha et al. 2012a). as compared to within-group differences. The equations
In order to maximize the relationship between the variables below give the expression of between- and within-column
under study, the factors are rotated. Factors rotation is used to variation:
facilitate interpretation by providing a simpler factor structure
2
(Sevilla et al. 2010). Factor rotation has been used to extract X
k X
j1
related variables and infer the processes that control surface TSS ¼ xij xþþ ¼ ðn 1Þs2 ð4Þ
water chemistry. Varimax rotation is applied to the PCs in order i1 ni
BSS=ðk 1Þ
One-way analysis of variance F statistics ¼ ð7Þ
WSS=ðn 1Þ
One-way analysis of variance (ANOVA) compares the An F ratio equal to or less than 1 indicates that there is no
means of more than two populations and it uncovers the significant difference between the groups and the null hy-
main and interaction effects of classification or independent pothesis is correct. However, if the F test proves the null
variables on a dependent variable. ANOVA uses the F hypothesis to be wrong, multiple comparison tests are used
statistics, which tests if the means of the groups formed by to further explore the specific relationships among different
one or more independent variables or combination of inde- groups.
pendent variables are significantly different. The F statistics
is based on the comparison of two estimates of variance: one
representing the variance within a group often refers to as
error variance and the other one representing the variance Results and discussion
due to differences in a group mean. If the two means do not
differ significantly, it can be concluded that group means Relationship between physicochemical and biological
come from the same sampling distribution of means and parameters of the water
there is no reason to claim that the group means differ.
However, if the group means differ, there is a reason to Tables 2, 3, and 4 present the results of the bivariate corre-
believe that they were drawn from different sampling lation coefficients for Jakara, Tsakama, and Gama-Kwari
Table 2 Pearson product moment correlation coefficient for Jakara River
4908
DO 1
BOD5 −0.592a 1
COD −0.443a 0.667a 1
SS 0.253 0.028 0.203 1
pH 0.338b 0.236 0.258 −000 1
NH3NL −0.659a 0.730a 0.656a −0.100 0.230 1
DS −0.203 0.195 0.479a −0.100 0.190 0.152 1
TS −0.197 0.196 0.486a −0.100 0.190 0.150 1.000a 1
NO3 −.540b 0.288b 0.309b 0.1200 0.100 0.446a −0.040 −0.040 1
Cl −0.182 0.183 0.473a −0.100 0.190 0.127 0.998a 0.997a −0.050 1
PO4 −0.635a 0.631a 0.608a −0.100 0.190 0.942a 0.128 0.127 0.671a 0.110 1
E. coli −0.622a 0.639a 0.529a −0.100 0.180 0.827a 0.132 0.131 0.541a 0.110 0.797a 1
FCB −0.611a 0.656a 0.511a −0.100 0.250 0.792a 0.250 0.249 0.297b 0.230 0.717a 0.922a 1
a
Correlation is significant at the 0.01 level (two-tailed)
b
Correlation is significant at the 0.05 level (two-tailed)
DO 1
BOD5 −0.686a 1
COD −0.644a 0.811a 1
SS 0.166 −0.040 −0.050 1
pH 0.140 0.099 0.043 0.355a 1
NH3NL −.0752a 0.895a 0.793a −0.000 0.100 1
DS −0.307b 0.187 0.165 −0.100 0.000 0.207 1
TS −0.305b 0.187 0.165 −0.100 0.000 0.206 1.000a 1
NO3 0.233 −0.330b −0.292b 0.050 0.100 −0.344b −0.239 −0.240 1
Cl −0.302b 0.181 0.169 −0.100 0.000 0.206 .998a 0.998a −0.242 1
PO4 −.672a 0.902a 0.741a −0.100 0.100 0.939a 0.208 0.210 −0.396a 0.210 1
E. coli −.444a 0.362a 0.330b −0.100 0.100 0.636a 0.185 0.190 −0.326b 0.210 0.628a 1
FCB −.515a 0.511a 0.451a 0.010 0.100 0.737a .327b 0.328b −0.365a 0.346a 0.757a 0.927a 1
a
Correlation is significant at the 0.01 level (two-tailed)
b
Correlation is significant at the 0.05 level (two-tailed)
Arab J Geosci (2013) 6:4903–4915
Arab J Geosci (2013) 6:4903–4915 4909
DO 1
BOD5 −0.363a 1
COD −0.426a 0.983a 1
SS −0.009 0.017 0.041 1
pH 0.185 0.022 0.040 −0.050 1
NH3NL −0.657a 0.056 0.123 0.110 −0.100 1
DS −0.462a 0.681a 0.679a 0.090 −0.050 0.336a 1
TS −0.462a 0.680a 0.678a 0.120 −0.050 0.337a 1.000a 1
NO3 −0.004 0.030 0.040 0.230 −0.470a 0.073 −0.000 −0.030 1
Cl −0.467a 0.694a 0.684a 0.080 −0.040 0.331a 0.994a 0.994a −0.040 1
PO4 −0.359a 0.030 0.042 0.140 −0.120 0.406a 0.080 0.078 −0.160 0.08 1
E. coli −0.622a 0.416a 0.479a −0.040 0.152 0.537a 0.370a 0.369a −0.030 0.376a 0.189 1
FCB −0.596a 0.504a 0.566a −0.060 0.276b 0.493a 0.444a 0.441a −0.030 0.453a 0.160 0.893a 1
a
Correlation is significant at the 0.01 level (two-tailed)
b
Correlation is significant at the 0.05 level (two-tailed)
Rivers respectively. The relationship between the 13 param- BOD5, COD, NH3NL had long been the basic means of
eters understudy was investigated using Pearson’s product– determining the degree of organic pollution in surface water.
moment correlation. The COD with BOD5 correlation can be explained by the
Preliminary analysis was conducted on the raw data fact that BOD5 represents liable organic matter which under-
before the analysis to ensure no violation of the assumption goes biotic decomposition and is part of COD. COD is a
of normality and linearity. The Levene test of homogeneity total organic matter content which is oxidized by a strong
of variances results shows that the variances are equal or oxidant (Cieszynska et al. 2012). The value of a correlation
homogeneous, because the Levene test obtained was very coefficient shows significant linear relationship between
small (0.132), and the corresponding p value was greater BOD5 and COD (r00.67) as well as other parameters in-
than the alpha value (p 0.05). cluding NH3NL (r00.73), PO4 (r00.63), E. coli (r00 .64),
As depicted in the tables, analysis of correlation coeffi- and FCB (r00.66) in the Jakara River. Further, the organic
cients shows inverse relationship for variables describing the compounds have higher linear correlation in the Tsakama
oxygen conditions of water and organic compound. The value River with BOD5 and COD (r00.81), BOD5 and NH3NL
of correlation coefficients equals −0.59, −0.69, and −0.37 for (r00.89), BOD5 and PO4 (r00.90), BOD5 and E. coli (r0
the pair DO and BOD5, −0.44, −0.64, and −0.43 for the pair 0.36), BOD5 and FCB (r00.51). Similarly, Gama-Kwari
DO and COD, −0.66, −0.75, and −0.66 for the pair DO and River revealed a positive linear correlation coefficient be-
NH3NL, −0.62, −0.44, and −0.62 for the pair E. coli, and tween BOD5 and COD (r00.98), E. coli (r00.41), FCB (r0
−0.61, 0.52, and 0.59 for the pair DO and FCB in Jakara, 0.50), Cl (r00.69), DS (r00.68), and TS (r00.68). These
Tsakama, and Gama-Kwari Rivers, respectively. parameters could be taken as representing the major sources
It is not surprising that the inverse relationship exists of river water quality variation. Many of the correlations can
between BOD5, COD, NH3NL, E. coli, and FCB with be explained by the organic compound used for domestic
DO. Lower level of concentration of DO in the rivers activities in the vicinity of the studied rivers. The present
may be due to the activities of microorganism in the finding seems to be consistent with other research which
water resulting in less penetration of light in the water found that urban land type together with the effects of
and subsequent decrease in photosynthesis activities increased population, domestic water use, and industrial
(Pejman et al. 2009; Palma et al. 2010). Higher rate wastewater were positively associated with increase in water
of bacterial decomposition of organic matter in the quality variations (Singh et al. 2004; Mustapha et al. 2012a).
surface water may increase the demand of oxygen for Large amount of biodegradable organic compound may
their metabolic activities and these would be expected affect surface water because organism in the water use the
to have exhibited a significant inverse correlation be- available DO to breakdown the waste. This process can
tween the parameters (Kumar and Reddy 2008). The deplete the supply of oxygen in the water needed by aquatic
correlations of organic biodegradable compounds life resulting in fish kill, odors, and overall degradation of
seemed to favor the growth of coliform bacteria. the water quality (Carpenter et al. 1998).
4910 Arab J Geosci (2013) 6:4903–4915
Surface water quality sources apportionment using principal Table 6 Varimax rotated component matrix Tsakama River
component analysis and factor analysis VF1 VF2 VF3 VF4
River water source apportionment is an important environ- BOD5 0.941 0.071 0.172 0.051
mentrics approach aiming to the estimation of contribution of COD 0.898 0.065 0.104 0.004
each parameter (Simeonov et al. 2003; Singh et al. 2004; NH3NL 0.859 0.064 0.456 0.069
Shrestha and Kazama 2007). PCA/FA assumes that the total PO4 0.820 0.063 0.492 0.064
from each element is made up of the sum of elemental con- DO −0.758 −0.204 −0.24 0.224
centrations from each identified polluting or natural cause TS 0.101 0.987 0.12 −0.022
component (Juahir et al. 2009). To obtain more reliable infor- DS 0.102 0.986 0.12 −0.036
mation about the relationships among the variables, PCA/FA Cl 0.094 0.983 0.141 −0.033
was used separately for each river to further explore the extent E. coli 0.231 0.045 0.939 0.014
of physiochemical and biological parameters and sources FCB 0.375 0.187 0.872 0.100
identification. Varimax rotation method was used to maximize NO3 −0.215 −0.187 −0.452 0.100
the sum of the variance of the factor coefficient, which better pH 0.060 −0.021 0.063 0.836
explained the possible group/sources that influenced the water SS −0.065 −0.037 −0.056 0.797
chemistry in the Jakara Basin (Singh et al. 2004). The factor Eigenvalue 5.72 2.63 1.42 1.09
loadings were ranked following the correlation coefficient Variance (%) 44.02 20.25 10.90 8.43
matrix between the variables. CV (%) 44.01 64.27 75.17 83.60
The calculated factor loadings, cumulated variance
explained by each factor in the PCA are presented in Tables 5, CV cumulative variance
6, and 7.
Preliminary analysis prior to factor analysis were al. 2003; Mustapha et al. 2012b). Four factors with
conducted to ensure that no violation of the assumptions eigenvalues greater than 1 were extracted from the
of Kaiser–Meyer–Olkin (KMO) measure on the sam- varimax-rotated factor analysis for each river water data.
pling adequacy and Bartlett test of sphericity. The Factor score represent the cumulative contribution of all
KMO result was 0.82, and the Bartlett sphericity test parameters loaded on a particular factor.
was significant (0.001, p<0.05), showing that PCA/FA The first varimax component (VF1) in the Jakara
could be considered appropriate and useful to provide River water sets explains more than 42 % of the total
significant reduction in the data dimensionality (Liu, et variance and is heavily loaded on NH 3NL, E. coli,
Table 5 Varimax rotated component matrix Jakara River Table 7 Varimax rotated component matrix Gama-Kwari River
a b
12.179478
12.179478
12.179478
12.179478
N N
12.123304
12.123304
12.123304
12.123304
Legend Legend
5 3
12.067130
12.067130
12.067130
12.067130
12.010956
12.010956
12.010956
12.010956
11.954782
11.954782
11.954782
11.954782
0 1.5 3 6 9 12 15 0 1.5 3 6 9 12 15
Kilometers Kilometers
8.454725 8.534440 8.614155 8.693870 8.773585 8.454725 8.534440 8.614155 8.693870 8.773585
8.454725 8.534440 8.614155 8.693870 8.773585 8.454725 8.534440 8.614155 8.693870 8.773585
c d
12.179478
12.179478
12.179478
12.179478
N N
12.123304
12.123304
12.123304
12.123304
Legend Legend
12.067130
12.067130
12.067130
12.067130
12.010956
12.010956
12.010956
12.010956
11.954782
11.954782
11.954782
11.954782
0 1.5 3 6 9 12 15 0 1.5 3 6 9 12 15
Kilometers Kilometers
8.454725 8.534440 8.614155 8.693870 8.773585 8.454725 8.534440 8.614155 8.693870 8.773585
8.454725 8.534440 8.614155 8.693870 8.773585 8.454725 8.534440 8.614155 8.693870 8.773585
e f
12.179478
12.179478
12.179478
12.179478
N N
12.123304
12.123304
12.123304
12.123304
Legend Legend
12.067130
12.067130
12.067130
12.067130
12.010956
12.010956
12.010956
12.010956
11.954782
11.954782
11.954782
11.954782
0 1.5 3 6 9 12 15 0 1.5 3 6 9 12 15
Kilometers Kilometers
8.454725 8.534440 8.614155 8.693870 8.773585 8.454725 8.534440 8.614155 8.693870 8.773585
g
12.179478
12.179478
N
12.123304
12.123304
Legend
12.067130
12.067130
12.010956
12.010956
11.954782
11.954782
0 1.5 3 6 9 12 15
Kilometers
8.454725 8.534440 8.614155 8.693870 8.773585
Fig. 3 Distribution of pollutants in Jakara, Tsakama, and Gama-Kwari Rivers a BOD5, b NH3NL, c COD, d DO, e E. coli, f FCB, and g PO4
Arab J Geosci (2013) 6:4903–4915 4913
Conclusions
Fig. 4 Mean difference between Jakara, Tsakama, and Gama-Kwari
Rivers In this study, surface water quality of Jakara, Tsa-
kama, and Gama-Kwari Rivers were investigated, and
combination of statistical techniques were used includ-
and Gama-Kwari Rivers. The distribution of the pollutants ing bivariate correlation, principal component analysis
were given in Fig. 3 and interpreted as suggested by Liu et (PCA) and one-way analysis of variance (ANOVA).
al. 2003 as high, moderate, and low corresponding to >50, Results from the bivariate correlation showed that the
0.51–0.74, and 0.75 and above factor loading. Figure 3 relationship among the variables understudy were
showed high concentration of BOD5, NH3NL, COD, PO4 helpful to identify the surface water quality pollution
with low concentration of E. coli, FCB and DO in Tsakama sources using PCA/FA. The PCA/FA of the studied
River, high distribution of NH3NL, E. coli, FCB, PO4, with water bodies revealed four independent varimax com-
moderate concentration of BOD5, COD and a low concen- ponents in each river water quality data sets explain-
tration of DO in River Jakara and high concentration of ing 84.88, 83.59, and 78.68 % of the variation in the
BOD5, NH3NL, COD with low distribution of E. coli, surface water quality for Jakara, Tsakama, and Gama-
FCB, DO, and PO4 in Gama-Kwari River. Kwari Rivers respectively. These components are re-
lated to organic compound from the anthropogenic
influences through direct dumping of waste water into
River water quality mean differences the surface water. The component which mostly
explained the variation in the data set had the stron-
A one-way between-groups analysis of variance was gest positive contribution of BOD 5 , COD, NH 3 NL,
performed to investigate water quality mean differences PO 4 , E. coli, FCB and a negative loading on the
obtained from Jakara, Tsakama, and Gama-Kwari Riv- available DO. The direct discharge of domestic efflu-
ers. The mean score obtained from the three rivers are ents and the carrying of surface run-off from the
depicted in Fig. 4. The null hypothesis states that there metropolitan are factors linked to the urban expansion
are no differences in the mean samples of river water and population growth which promotes the degradation
quality between the three rivers under study. Preliminary of the water bodies. The result form the ANOVA
assumption testing was conducted to check for showed that all the three rivers had the same pollution
status. The study recommended that the implementa-
tion of a system for sewage collection and treatment
before disposal, as well as structural and non-
Table 8 One-way ANOVA results structural measures which will decrease surface run-
off are examples of interventions to recuperate and
Sum of Mean square F p value preserve the quality of the surface water resources in
squares
the Jakara Basin.
Between-groups 1.64E+09 8.21E+08 1.563 0.223
Within-groups 1.89E+10 5.25E+08
Total 2.06E+10 Acknowledgments We will like to thank the anonymous reviewers
for their valuable comments on the manuscript.
4914 Arab J Geosci (2013) 6:4903–4915
Taiwo A, Olujimi O, Bamgbose O, Arowolo T (2010) Surface water Grombalia phreatic aquifer (Northeastern Tunisia). Arab J Geosci.
quality monitoring in Nigeria: a situational analysis and future doi:10.1007/s12517-012-0617-3
management strategy. J Sustain Manag 2:301–320 Vidal M, Lopez P, Santoalla M, Valles V (2000) Factor analysis for the
Tiri A, Boudoukha A (2010) Hydrochemical analysis and assessment study of water resources contamination due to the use of livestock
of surface water quality in Koudiat Medour Reservoir, Algeria. slurries as fertilizer. Agric Water Manag 45:1–15
Eur J Sci Res 41:273–285 Yang L, Song X, Zhang Y, Yuan R, Ma Y, Han D, Bu H (2012) A
Tlili-Zrelli B, Hamzaoui-Azaza F, Gueddari M, Bouhhlila R (2012) hydrochemical framework and water quality assessment of river
Geochemistry and quality assessment of groundwater using water in the upper reaches of the Huai River Basin, China.
graphical and multivariate statistical methods. A case study: Environ Earth Sci. doi:10.1007/s12665-012-1654-7