You are on page 1of 15

Environ Earth Sci (2016) 75:21

DOI 10.1007/s12665-015-4779-7

ORIGINAL ARTICLE

Assessment of groundwater quality in Yaoundé area, Cameroon,


using geostatistical and statistical approaches
William Assatse Teikeu1 • Jorelle Larissa Meli’i1 • Philippe Njandjock Nouck1 •

Tabod Charles Tabod1,2 • Francoise Enyegue A Nyam1 • Zakari Aretouyap1

Received: 13 February 2015 / Accepted: 8 July 2015


Ó Springer-Verlag Berlin Heidelberg 2015

Abstract This paper presents the results of a survey Keywords Groundwater quality  Geostatistical 
carried out in 2010 and 2012 aimed at evaluating the Principal component analysis  Yaoundé
quality of the groundwater resources as part of emergency
program in drinking water of the Yaoundé area, Cameroon.
The groundwater quality parameters were analyzed from Introduction
the 42 bore wells of the region, and the thematic maps were
generated using geostatistical and statistical concepts. Groundwater is an important source of drinking water for
Ordinary Kriging were used as geostatistical approach for many people around the world, especially in urban areas
preparation of thematic maps of the groundwater quality (Teikeu et al. 2012a, b). Groundwater can be contaminated
parameters such as electrical conductivity (EC), total dis- from natural sources or numerous types of human activities
solved solid (TDS), bicarbonate, chloride, calcium, mag- (Yongsi Nguendo 2011). In Cameroon, the authorities have
nesium, iron, sodium, and potassium. Experimental favored the availability of water at the expense of its
semivariogram values were fitted well in spherical model quality. Analyzes are targeting only a few physicochemical
for the water quality parameters, such as TDS, EC, bicar- elements, and it is only when the water is cloudy or has
bonate, chloride, magnesium, sodium, and iron, and in abnormal or even the taste is unusual that specific elements
exponential model for PH. Three factors were extracted by are determined. Contamination of groundwater can result
principal component analysis, which explains 91.82 % of in poor drinking water quality and potential health prob-
the total variation. The first factor captures variables of EC, lems (Djebebe-Ndjiguim et al. 2013; Tamutou et al. 2013,
TDS, Cl-, Ca2?, Mg2?, Fe2?, Na?, K?, and TH. The Parmar and Bhardwaj 2014). Geostatistical approach is
second factor is significantly related to EC, TDS, and widely used by many researchers for the analysis of spatial
HCO 3 , while PH accounted for the greatest loading for variations of groundwater characteristics (Delhomme 1979;
factor 3. Water Quality Index showed that 74.58 % of the Adhikary et al. 2010). The spatial distribution of polluted
samples were good for drinking, and the remaining were groundwater show some heterogeneity, and the pollutant
moderate and poor for drinking purpose. In fact, risk concentration values are rarely available for every possible
assessment of groundwater quality is useful for proper location of an area. The measurement of pollutant con-
management of groundwater resources for minimizing the centration at every location is not always feasible in view
pollution threat. of the time and the cost involved in data collection.
Therefore, prediction of values at other locations based
upon selectively measured values could be one of the
& Philippe Njandjock Nouck alternatives. In this context, to predict the concentration of
pnnouck@yahoo.com
pollutants at unmeasured locations, the geostatistical
1
Laboratory of Geophysics, University of Yaounde I, techniques can be used. The basic assumption in using
812, Yaoundé, Cameroon geostatistics is that the properties in the earth have some
2
Department of Physics, University of Bamenda, 39, Bambili, spatial continuity up to a certain lag distance (Oyoa et al.
Cameroon 2012). In geostatistics, the Kriging interpolation is

123
21 Page 2 of 15 Environ Earth Sci (2016) 75:21

regarded as the optimal interpolation and is the most heterogeneous. The stream and river network is dense and
widely used. It estimates an unknown point by making use arborescent. The city is indeed drained by the Mfoundi
of known points (Uyan and Cay 2013). Geostatistical River and its many tributaries. The hydrodynamic func-
results using Kriging techniques are more efficient when tioning of the coupled rock/soil system acts as two-layered
data for variables are distributed normally (Wu et al. 2011). systems in which weathered zone has a capacitive role and
The main aims of this investigation are to provide an fractures have a first-rate drainage role (Djeuda Tchapnga
overview of present groundwater quality for parameters et al. 1999).
such as calcium, magnesium, iron, sodium, potassium,
bicarbonate, chloride, electrical conductivity (EC), pH,
total dissolved solid (TDS), and total hardness (TH) and to Materials and methods
realize the groundwater quality map in Yaoundé area,
Cameroon, using geostatistical and statistical techniques. Samples collection and laboratory analysis

Study area A total of 42 groundwater samples were collected between


March 2010 and July 2012 (Table 1). Spectrophotometric
Yaoundé is located between latitudes 3°–5°N and longi- method was used to analyze for cations and anions. The
tudes 11°–13°E, over a mean altitude of 710.8 m (Fig. 1). physical parameter PH and electrical conductivity were
Annual rainfall is around 1626 mm and annual mean determined on the field using a calibrated PH meter and
temperature is 24.17 °C. Yaoundé counts four seasons conductivity meter, respectively.
including two (major and minor) rainy seasons and two
(major and minor) dry seasons. Due to the development of Factor analysis
the city, the natural vegetation is quickly degrading,
resulting in the degradation and erosion of the soils. The In the last two decades, factor analysis (FA) has been
area is covered by well-drained ferralitic soils on the hill successfully used to sort out the hydrochemical processes
slopes and by colluvial deposits in the valley. The base- and relationships of analyzed groundwater data (Amadi
ment consists of migmatites and gneisses (Mvondo et al. 2011). The purpose of factor analysis is to simplify the
2007). It is neither porous nor soluble, but it is the dis- quantitative description of a system by determining the
continuities (faults, diaclases) that give fissure permeability minimum number of new variables necessary to reproduce
to the formation; this last being anisotropic and various attributes of the data (Amadi et al. 2012). This

Fig. 1 Location of the study area and the sampling bore wells

123
Environ Earth Sci (2016) 75:21 Page 3 of 15 21

Table 1 Chemical composition of groundwater samples collected from the study area
Samples T PH EC (ls/ TDS TH Ca2? Na? Mg2? K? Fe2? Cl- HCO 3
number (°C) cm) (mg/l) (°F) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l)

W1 24 7.23 191.3 33.6 0.95 2.57 1.07 0.71 0.21 0.03 12.11 122
W2 25 5.5 156.7 126.1 0.78 2.19 0.88 0.58 0.25 0.08 9.92 73.2
W3 25 6.5 510 245 2.5 7.29 2.85 1.89 0.82 0.13 32.28 292.6
W4 25 6.5 267 110.1 1.4 4.01 1.61 1.06 0.46 0.15 18.17 73.3
W5 25 6.53 133.5 58.6 0.66 1.86 0.75 0.49 0.22 0.09 8.45 73.2
W6 24.2 6.67 473 233 2.3 6.6 2.66 1.7 0.76 0.17 29.94 122
W7 23.9 6.96 321 158 1.59 4.48 1.8 1.19 0.52 0.05 20.32 175
W8 25 6.5 110.8 52.5 0.55 1.55 0.62 0.41 0.17 0.09 7.01 73.2
W9 25 6.65 244 101.1 1.21 3.41 1.37 0.9 0.39 0.18 15.45 97.6
W10 25 6.5 395 151.1 1.96 5.52 2.22 1.47 0.64 0.49 25 195.2
W11 25 6.5 318 126.1 1.88 4.44 1.79 1.18 0.51 0.45 20.13 170.8
W12 23 7.52 368 17.9 0.18 0.51 0.21 0.13 0.056 0 2.33 73.2
W13 25 6.5 382 191 14 41.89 16.85 11.17 4.8 0 170 170.8
W14 24.7 6.54 135.5 67.75 7.2 20.79 8.26 5.54 2.41 0.56 94.26 97.5
W15 24 6.52 117.4 57 0.58 1.64 0.66 0.43 0.19 0.17 7.43 58.56
W16 25 6.54 383 192 15 41.8 16.85 11.17 4.86 0.18 179 258.64
W17 25 6.65 67.1 35.55 1.5 4.19 1.68 1.11 0.48 0 19 48.8
W18 23.5 6.6 325 162.5 22.41 62.81 25.27 16.75 7.29 0.92 182 271.81
W19 23.4 6.76 218 109 22.48 62.84 25.29 16.77 7.39 0.3 195 146.4
W20 24.8 6.53 71.9 35.95 4.5 11.87 4.77 3.16 1.38 0 53.81 48.8
W21 23.4 6.87 83.5 41.75 4.5 12.85 5.17 3.4 1.49 0.03 58.24 74.66
W22 24.6 6.53 186.4 93.2 9.4 26.53 10.67 7.07 3.08 0.78 120.28 122
W23 22.6 6.66 170 40.5 8 16.7 6.74 4.47 1.94 0.02 78 97.6
W24 25 6.56 83.4 41.7 0.4 1.16 0.47 0.31 0.13 0.03 5.28 48.8
W25 25 6.57 266 133 9.9 27.93 11.23 7.44 3.24 0.07 20 219.6
W26 25 6.65 125.8 58.2 0.62 1.76 0.71 0.47 0.20 0.29 7.96 73.2
W27 24.2 6.59 116.3 65 0.78 5.24 0.73 0.67 0.24 0.19 8.1 69
W28 24.5 6.45 117.9 59 0.49 2.34 0.54 0.33 0.17 0.24 7.74 56.27
W29 24.8 6.58 144.1 66.8 0.63 1.63 0.69 0.98 0.24 0.11 7.89 61.4
W30 24.7 6.52 113.29 71 0.59 3.5 0.8 0.47 0.13 0.14 7.24 53.86
W31 23.9 6.58 429 173.8 24.67 69.54 26.48 19.81 7.83 0.82 193 289.29
W32 25 6.69 129.3 51.8 1.63 1.89 0.83 0.97 0.78 0.34 8.63 61
W33 25 6.57 168 54.2 1.49 1.43 0.99 0.85 0.67 0.29 7.51 78
W34 25 6.51 149 51.9 1.34 1.29 0.86 0.87 0.56 0.3 7.54 72
W35 24.8 6.55 139.8 69.81 7.9 24.64 8.38 7.91 2.62 0.61 95.32 93.2
W36 24.7 6.49 132.4 62.93 7.7 24.73 8.17 7.18 2.19 0.57 92.69 91.3
W37 23.5 6.87 331 172 1.62 4.35 1.62 1.25 0.38 0.04 20.89 168
W38 23.7 6.81 335 176 1.62 4.97 1.55 1.21 0.34 0.03 21.23 166
W39 25 6.54 136.9 56.9 0.79 4.96 0.78 1.43 0.32 0.1 0.34 73.9
W40 25 6.38 131.3 53.4 0.86 4.48 0.75 1.35 0.39 0.08 0.32 71.2
W41 25 6.47 135.4 55.2 0.78 4.87 0.73 1.49 0.38 0.09 0.35 72.4
W42 25 6.54 263 139 9.85 29.81 11.82 7.72 3.58 0.08 28.1 214.7

approach reduces data into a small set without losing the This step helps in increasing the influence of variables
key information from the original data. In order to carry out having small variance and reducing the influence of the
the factor analysis, the raw data should be standardized. variables having large variance. Then the correlation

123
21 Page 4 of 15 Environ Earth Sci (2016) 75:21

coefficients are evaluated which will be helpful in All parameters of groundwater quality have best-fitted
explaining the structure of the underlying system which variogram of spheric and exponential model (Fig. 2).
produced the data (Uddameri et al. 2014). The degree of Table 2 illustrates parameters of groundwater quality
mutually shared variability between individual pairs of variograms.
water quality parameters can be represented by correlation The validation and the accuracy of the developed vari-
coefficient. The correlation coefficient is given as, ogram model can be tested via a technique called cross-
Pn validation. This will verify the adequacy between the data
i¼1 ðxm  xi Þðym  yi Þ
rx;y ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn 2 Pn 2 ð1Þ and the variogram model used. It involves estimating each
i¼1 ðxm  xi Þ i¼1 ðym  y€i Þ data whose value is known by eliminating data entry
(Kyriakidis 2004). Comparing between estimated value
In this expression the correlation coefficients (rx,y) is sim-
and observed value is a cloud of points whose configura-
ply the sum (over all samples) of the products of the
tion around the bisector certifies the robustness of the
deviations of the x-measurements and the y-measurements
model: The higher the point cloud is tightened around the
on each sample, from the mean values of x and y, respec-
bisecting line, the better estimate values by Kriging
tively, for the complete set of samples (Chen-Wuing et al.
(Fig. 3). One of the main advantages of Kriging, although
2003). The eigenvalues and factor loadings for the corre-
there is no general consensus on its usefulness, is that it
lation matrix are determined, and scree plot is drawn. One
presents the possibility of interpolation estimation error of
of the most important steps in factor analysis is the
the regionalized variable where there are no initial mea-
extraction of factors based on the variances and covari-
surements. This feature offers a measure of the estimation
ances of the variables. The eigenvalues and eigenvectors
precision and reliability of the spatial variable distribution.
are evaluated which represent the amount of variance
The general equation of the Kriging method is as follows:
explained by each factor. The factors with eigenvalues
greater than one depict more variation in data than indi- X
n
Z  ðxp Þ ¼ ki Zðxi Þ ð3Þ
vidual variable. Finally, by the process of rotation, the i¼1
loading of each variable on one of the extracted factors is
maximized and the loadings on all the other factors are In order to achieve unbiased estimations in ordinary
minimized. These factor loadings are useful in grouping the Kriging, the following set of equations should be solved
water quality parameters and providing information for simultaneously.
interpreting the data. The XLSTAT 2014 software package X n
ki cðxi ; xj Þ  l ¼ cðxi ; xj Þ ð4Þ
and the MS-EXCEL sheets were used for the statistical
i¼1
analysis.
X
n
ki ¼ 1 ð5Þ
Geostatistical approach i¼1

Here, Z  ðxp Þ is the kriged value at location xp , Zðxi Þ is the


The semivariogram is a fundamental tool in geostatistics.
known value at location xi , ki is the weight associated with
The empirical semivariogram c (h) is defined as half the
the data, l is the Lagrange multiplier, and cðxi ; xj Þ is the
average quadratic difference between two observations of a
value of variogram model corresponding to a vector with
variable separated by a distance vector h (Njandjock et al.
origin in xi and extremity in xj .
2013). It is calculated according to the following formula:

1 X
NðhÞ
Water Quality Index (WQI)
YðhÞ ¼ ½Z ðxi Þ  Zðxi þ hÞ2 ð2Þ
2N i¼1
WQI indicates the quality of water in terms of index
Here, c (h) means the semivariogram value at the distance, number which represents overall quality of water for any
h; N (h) means the total number of the variable pairs sep- intended use. It defined as a technique of rating that pro-
arated this distance, and Z (x) means the value of the vides the composite influence of individual water quality
variable. Before the geostatistical estimation, a variogram parameter on the overall quality of water (Sadat-Noori
model is calculated for classes of distance between sample et al. 2014). The indices are among the most effective ways
pairs. The most widely used models are spherical, expo- to communicate the information on water quality trends to
nential, gaussian, cubic, and logarithmic (Meli’i et al. the general public and in water quality management
2013). Geostatistical program (Autogrig) developed by the (Nshagali et al. 2014). In the formulation of water quality
Geophysical Laboratory of the University of Yaoundé I in index the relative importance of various parameters
a MATLAB platform was used. depends on intended use of water. Mostly it is done from

123
Environ Earth Sci (2016) 75:21 Page 5 of 15 21

Fig. 2 Best-fitted semivariograms model of groundwater quality concentration, i Fe2? concentration, j Cl- concentration, k HCO
3
parameters in the study area. a PH, b EC, c TDS, d TH, e Ca2? concentration, l Water Quality Index
concentration, f Na? concentration, g Mg2? concentration, h K?

123
21 Page 6 of 15 Environ Earth Sci (2016) 75:21

Fig. 2 continued

Table 2 Summary of best-


Ground water parameter Best-fitted model Nugget Sill Range RMS
fitted models for different
CO C ? CO a
groundwater quality parameter
PH Exponential 0.0084 0.0647 0.0127 7
EC Spherical 41.3595 20,701.3000 0.0380 3
TDS Spherical 2.7370 4811.62 0.0443 2
TH Spherical 0.0052 53.9521 0.0559 4
2?
Ca Spherical 0.7582 460.098 0.0445 5
Na? Spherical 5.4612 43.9488 0.04759 5
Mg2? Spherical 0.3461 31.7011 0.05363 4
K? Spherical 0.0019 5.4510 0.0494 6
Fe2? Spherical 0.0008 0.0752 0.0441 2
HCO 3 Spherical 6.8174 6792.47 0.0361 3
Cl- Spherical 1.5845 4864.05 0.0388 5
WQI Spherical 8.1425 7122.38 0.0484 2

the point of view of its suitability for human consumption. where Ci is the observed concentration of the ith water
WQI was calculated using the weighted arithmetic index quality parameter; Si is the concentration desirable limit of
method. The quality rating scale for each parameter qi was the ith water quality parameter and Cmin is the minimum
calculated by using this expression: concentration of the parameter reflecting best water
 
Ci  Cmin quality.
qi ¼  100 ð6Þ
Si  Cmin

123
Environ Earth Sci (2016) 75:21 Page 7 of 15 21

123
21 Page 8 of 15 Environ Earth Sci (2016) 75:21

b Fig. 3 Scatter plot of estimated versus observed values of ground- Table 3 Standard limit concentration and weight of groundwater
water quality parameters: a PH, b EC, c TDS, d TH, e Ca2? quality parameters
concentration, f Na? concentration, g Mg2? concentration, h K?
concentration, i Fe2? concentration, j Cl- concentration, k HCO Groundwater quality Standard limits concentration Weight
3
concentration, l Water Quality Index parameters (WHO 2004) (Wn)

PH 6.5–8.5 0.0345
The minimum value for all the parameters considered in
EC 300 0.0009
the model was 0 except pH (for pH 6.5–8.5 represent best
TDS 500 0.0006
water quality).
TH 35 0.0079
The unit weight (Wi) to various water quality parameters
Ca2? 75 0.0037
is inversely proportional to the recommended standards for
Na? 200 0.0012
the corresponding parameters.
Mg2? 30 0.0092
K
Wi ¼ ð7Þ K? 12 0.023
Si Fe2? 0.3 0.92
where K is the constant given by this expression. HCO
3 240 0.0012
Cl- 200 0.0014
1
K¼  ð8Þ Total weight 1.0016
1
S1 þ S12 þ    þ S1n

The overall WQI was calculated by the following equation:


Pn Results and discussion
qi w i
WQI ¼ Pi¼1
n ð9Þ
i¼1 wi In the following, the 12 groundwater quality parameters, T,
Concentration limit value and weight of ith water PH, EC, TDS, TH, Ca2?, Na?, Mg2?, K?, Fe2?, HCO 3,
quality parameters are listed in Table 3. Cl- will be analyzed according to their statistical measures

Fig. 3 continued

123
Environ Earth Sci (2016) 75:21 Page 9 of 15 21

Table 4 Statistical summary of drinking water quality in Yaoundé area


Parameters T (°c) PH EC (ls/ TDS TH Ca2? Na? Mg2? K? Fe2? HCO 3 Cl-
cm) (mg/l) (°F) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l)

Max 25 7.52 510 245 24.67 69.54 26.48 19.81 7.83 0.92 292.6 195
Min 22.6 5.5 67.1 17.9 0.18 0.51 0.21 0.13 0.056 0 48.8 0.32
Mean 24.52 6.59 216.07 96.45 4.74 13.54 5.19 3.70 1.54 0.22 118.33 45.19
SD 0.66 0.27 119.64 59.19 6.44 18.17 7.23 4.96 2.09 0.24 70.61 59.32
CV (%) 2.69 4.12 55.37 61.37 135.76 134.17 139.17 134.04 135.80 108.08 59.67 131.25

such as minimum, maximum, mean, and standard devia- and PH, respectively, and then they are elevated to the TH
tion. The results are presented in Table 4. (135.76), Ca2? (134.17), Na? (139.17), Mg2? (134.04), K?
(135.80), Fe2? (108.08), and Cl- (131.25) (Table 4). These
Physicochemical parameters results show that the pH and the temperature are thus very
homogeneous and vary very little from one point to
The groundwater temperature is between 22.6 and 25 °C another, but TH, Ca2?, Na?, Mg2?, K?, Fe2?, and Cl- are
for all bore wells, a range close to the annual mean air therefore widely heterogeneous in space.
temperature in the Yaoundé region of around 24.17 °C.
The pH of the groundwater is acidic and ranges between Major ions
5.5 and 7.5, as expected considering the clayey, siliceous
and medium carbonate nature of the aquifer matrix. The relative tendency of major ions in groundwater is in
Although pH usually has no direct impact on consumers, it the following order Ca2? C Na? C Mg2? C K? C Fe2?
is one of the most important operational water quality and HCO -
3 C Cl . Ca
2?
concentration ranged from 0.59
parameters, with the optimum pH required often being in to 69.54 mg/l in Yaoundé groundwater. Knowing that the
the range of 6.5–8.5 (WHO 2004). Spatial distributions of maximum permissible limit of calcium concentration set
concentrations of pH are shown in Fig. 4a. It is shown that, by the WHO for drinking water was specified as 75 mg/l
PH of the acid groundwater occurs throughout the study (Table 4), it is observed that all of groundwater and
area. Yaoundé aquifer samples show lower concentration than
Electrical conductivity (EC) shows a wide range of the permissible limit (Fig. 4e). The concentration of
values, from 67.1 to 510 ls/cm. As shown in Fig. 4b, the sodium Na? varies to 0.21 and 26.48 mg/l in the
water samples has an electrical conductivity between 80 groundwater. According to the WHO guidelines, the
and 250 ls/cm (40.87 %), class corresponding to weakly maximum admissible limit for drinking water is 200 mg/
mineralized waters against 59.13 % corresponding to l. In the study area, all groundwater samples do not
medium mineralized waters (250480 ls/cm). Indeed, the exceed the maximum permissible limit for drinking
full mineralization of dissolved salts in the water (Kuitcha water (Fig. 4f). The magnesium concentration Mg2? of
et al. 2013). the studied samples ranged from 0.13 to 19.81 mg/l.
The estimated total dissolved solids (TDS) ranged from According to the standard, all the Mg2? concentrations
17.9 to 245 mg/l in bore wells. The variation in the value in the study area are within the maximum permissible
TDS (Fig. 4c) reflects the changes in water level and limit (Fig. 4g). The potassium ion concentration K?
leaching of minerals into groundwater. Although the ranged from 0.06 to 7.83 mg/l with mean value of
influence of anthropogenic component is minimal, TDS 1.54 mg/l; all of the samples were within the permissible
observation is made based on the fact that migrations of limit (Fig. 4h). Iron concentration Fe2? ranged from 0 to
dissolved solids in hard rock areas are slower when com- 0.92 mg/l in bore wells and exceeds the acceptable limit
pared to other terrains. of 0.3 mg/l in samples W18, W19, W22, W31, W32,
Total hardness (TH) is caused by dissolved calcium and, W35, and W36. It can be seen in Fig. 4i that the iron
to a lesser extent, magnesium. The minimum and maxi- concentration is permissible limit from south to north of
mum of the total hardness values are, respectively, 0.18 the study area and that it exceeds maximum contaminant
and 24.67 °F. Total hardness from north, southwest, east, level 0.3 mg/l northwest of the city. Chloride Cl- con-
and northwest of study area was estimated to be \8 °F centration ranged from 0.32 to 195 mg/l. There was no
(Fig. 4d). Groundwater in the area is soft. water bore well in which the chloride concentration Cl-
The coefficient variation (CV) values of the physical exceeds the permissible limited. As indicated by Fig. 4j,
parameters are very low (2.70 and 4.13) for the temperature chloride concentration decreased from northwest to

123
21 Page 10 of 15 Environ Earth Sci (2016) 75:21

Fig. 4 Thematic maps of groundwater quality parameters in the study area. a PH, b EC, c TDS, d TH, e Ca2? concentration, f Na?
concentration, g Mg2? concentration, h K? concentration, i Fe2? concentration, j Cl- concentration, k HCO
3 concentration

southeast. In a wide area around the south and north part acceptable limit of 240 mg/l (WHO 2004) in samples
of the study area, \60 mg/l chloride concentration W3, W16, W18, and W31. It can be seen in Fig. 3k that
occurs. The content in bicarbonate concentration HCO 3 the HCO 3 concentration exceeds permissible limit in the
ranged from 48.8 to 292.6 mg/l and exceeds the west side of the studied region.

123
Environ Earth Sci (2016) 75:21 Page 11 of 15 21

Fig. 4 continued

Factor analysis this analysis, the labels strong, moderate, and weak refer to
correlation values [0.75, 0.5–0.75, and 0.3–0.5, respec-
The Pearson product moment correlation coefficients for tively (Chen-Wuing et al. 2003). It is noted that, the worthy
42 samples are presented in Table 5. For the purposes of water mineralization represented by electrical conductivity

123
21 Page 12 of 15 Environ Earth Sci (2016) 75:21

Fig. 4 continued

is determined mainly by the total dissolved solids and (r = 0.994), Ca2? (r = 0.996), Cl- (r = 0.937), Fe2?
bicarbonate concentration (r [ 0.8). Calcium and magne- (r = 0.543) and HCO 3 (r = 0.642).
sium were well correlated with bicarbonates (r [ 0.6) and Factor analysis was performed to identify characteristics
chloride (r [ 0.9). Total hardness is highly bound to major of water quality variables in all stations (Tanriverdi et al.
ions as follows: K? (r = 0.997), Na? (r = 0.998), Mg2? 2010), and the factor loading matrix is listed in Table 6.

123
Environ Earth Sci (2016) 75:21 Page 13 of 15 21

Table 5 Pearson correlation coefficients of chemical parameters


Chemical variables PH EC TDS TH Ca2? Na? Mg2? K? Fe2? HCO
3 Cl-

PH 1.000
EC 0.205 1.000
TDS -0.138 0.867 1.000
TH -0.011 0.350 0.392 1.000
Ca2? -0.023 0.345 0.399 0.996 1.000
Na? -0.013 0.359 0.410 0.998 0.997 1.000
Mg2? -0.024 0.340 0.387 0.994 0.998 0.993 1.000
K? -0.019 0.347 0.396 0.997 0.995 0.998 0.993 1.000
2?
Fe -0.148 0.142 0.152 0.542 0.543 0.529 0.563 0.540 1.000
HCO
3 0.061 0.808 0.813 0.642 0.640 0.649 0.632 0.640 0.339 1.000
Cl- -0.011 0.333 0.384 0.937 0.934 0.936 0.933 0.929 0.548 0.553 1.000

Three significant factors (eigenvalue[1) were extracted by Table 6 Factor analysis-rotated component matrix–varimax with
factor analysis, which explains 91.40 % of the total vari- Kaiser normalization
ance. Factor F1 expressed 64.76 % of the variance, is the
Component PC1 PC2 PC3
most important of all, followed by the factors F2 and F3
which expresses 16.83 and 9.81 % of the variance, PH -0.018 0.186 0.964
respectively. These factorial axes selected for this statisti- EC 0.522 0.815 0.055
cal analysis are representative of the variance of data set. TDS 0.564 0.746 -0.279
The factorial F1–F2 and F1–F3 cumulatively of the vari- TH 0.974 -0.189 0.061
ances are equal to 81.59 and 74.57 %, respectively. Given Ca2? 0.974 -0.189 0.048
these percentages expressed, we can now say that, the Na? 0.976 -0.174 0.057
mechanism that controls the chemical evolution of water in Mg2? 0.972 -0.202 0.045
the region can be described using the first two factors. K? 0.972 -0.189 0.052
Thus, our analysis focuses only on the F1–F2 factorial Fe2? 0.584 -0.276 -0.227
design that highlights the general trend. HCO
3 0.769 0.549 -0.034
The first factor (F1) accounts for 64.76 % of the total Cl- 0.928 -0.201 0.051
variance and has strong positive loadings on TH, Ca2, Na?, Eigen values 7.123 1.851 1.079
Mg2?, K?, HCO -
3 , and Cl , and a moderate loading of EC, % of variance 64.758 16.829 9.809
2?
TDS, and Fe (Fig. 5). The high correlation between these Cumulative% 64.921 81.588 91.397
variables and their proximity in the community circle
shows that the dissolution of these ions can be controlled
by the hydrolysis phenomenon and the contact of the water
with the surrounding geological formations (Kortatsi et al. factor loadings in factor 2, which indicates that they may
2008; Kuitcha et al. 2013). The second factor (F2) accounts be derived from rock–water interaction processes.
for 16.83 % of the total variance and is significantly related
to EC, TDS, and HCO 3 . Factor 2 opposes EC, TDS, and Groundwater quality map
HCO 3 with Cl -
, Ca 2?
, Mg2?, Fe2?, Na?, K?, and TH
contents. This axis, basically representing the calco-car- Figure 6 shows the final water quality map that was pro-
bonic parameters, describes factor dilution of groundwater duced by overlapping of the thematic maps as a result of
by water recharge and/or mineralization by water–soil/rock geostatistical analysis. The result of groundwater vulnera-
interaction (Alaya et al. 2014). The opposite evolution of bility to pollution assessment shows index values which
Ca2? and HCO 3 in factor 2 indicates that the increase in ranged from 20 to 180. The Water Quality Index was
HCO 3 content in groundwater is essentially related to reclassified into four classes which describe the quality of
cation exchange processes. The positive loadings of pH groundwater in the study region. According to the results of
values suggested that the major ion concentration is con- the groundwater vulnerability assessment, the study area
trolled by pH variations in the study area, whereas pH as a has been divided into four types of zones: excellent, good,
controlling factor rather than EC, TDS, and HCO 3 has low moderate, and poor water quality. The ranges and class of

123
21 Page 14 of 15 Environ Earth Sci (2016) 75:21

Table 7 Water quality classification scheme based on WQI value


Water Quality Index Water quality Water samples
(WQI) class (%)

\50 Excellent 24.81


50–100 Good 49.77
100–150 Moderate 17.10
[150 Poor 8.31

especially at Ahala, Odza, and Mendong. The good water


quality could be found on a large area at the north and
center of the area. It represents 49.77 % of total area. The
rest of the study area, which is about 25.41 %, has water
classified as moderate and poor quality levels. These phe-
nomena result of iron and bicarbonate concentration
especially.

Fig. 5 Representation of the principal components of the variables


measured in field samples Conclusion

In this study, an attempt has been made to investigate


groundwater quality status by the combined use of the GIS
and geostatistical techniques applied on a set of original
data. Geostatistical techniques create surfaces incorporat-
ing the statistical properties of the measured data. Principal
component analysis was utilized to identify the potential
source and evolution of groundwater of the region using an
extensive multivariate dataset. The results of the study
indicated certain general trends. In total, 91.42 % of the
observed variance was explained by the first three factors
in the groundwater. These factors suggest that the
groundwater quality ranged from weakly to moderately
mineralized. WQI was developed from the thematic maps
of generated for water quality parameters using overlay and
index method. WQI map clearly reveals the suitability of
groundwater quality for drinking purposes. As the iron and
bicarbonate concentrations exceed permissible limit in the
studied area, this study has shown that the deterioration of
groundwater quality in the study area can become a very
serious problem.

Acknowledgments The authors are grateful to the Ministry of


Scientific Research and Innovation, to the Ministry of Water Resource
and Energy, and to the anonymous reviewer who have permitted them
to conduct and perform this work. They also want to thank Pascal
ASSATSE, Justine MAGUEGUIM, the late Julienne Ngo MATIP,
and Jeannot NOUCK for their support.
Fig. 6 Final groundwater quality map

References
the groundwater quality index of WQI map are given
below in Table 7. The area for excellent water quality can Adhikary PP, Chandrasekharan H, Chakraborty D, Kamble K (2010)
be found at southwest of the region. These areas for Assessment of groundwater pollution in West Delhi, India using
geostatistical approach. Environ Monit Assess 167:599–615
excellent water quality cover 24.81 % of total area,

123
Environ Earth Sci (2016) 75:21 Page 15 of 15 21

Alaya MB, Saidi S, Zemni T, Zargouni F (2014) Suitability Nshagali BG, Njandjock NP, Meli’i JL, Arétouyap Z, Manguelle-
assessment of deep groundwater for drinking and irrigation use Dicoum E (2014) High iron concentration and pH change
in the Djeffara aquifers (Northern Gabes, south-eastern Tunisia). detected using statistics and geostatistics in crystalline basement
Environ Earth Sci 71:3387–3421 equatorial region. Environ Earth Sci. doi:10.1007/s12665-014-
Amadi AN (2011) Assessing the effects of Aladimma dumpsite on 3893-2
soil and groundwater using Water Quality Index and factor Oyoa V, Njandjock NP, Bisso D, Diab AD, Manguelé-Dicoum E
analysis. Aust J Basic Appl Sci 5(11):763–770 (2012) A geostatistical approach to map gravity data over
Amadi AN, Olasehinde PI, Yisa J, Okosun EA, Nwankwoala HO, Logone-Birni sedimentary basin (Chad-Cameroon). Eur J Sci
Alkali YB (2012) Geostatistical assessment of groundwater Res 93(2):183–189
quality from coastal aquifers of Eastern Niger Delta, Nigeria. Parmar KS, Bhardwaj R (2014) Water quality management using
Geosciences 2(3):51–59 statistical and time-series prediction model. Appl Water Sci.
Chen-Wuing L, Kao-Hung L, Yi-Ming K (2003) Application of factor doi:10.1007/s13201-014-0159-9
analysis in the assessment of groundwater quality in a blackfoot Sadat-Noori MS, Ebrahimi K, Liaghat MA (2014) Groundwater
disease area in Taiwan. Sci Total Environ 313:77–89 quality assessment using the Water Quality Index and GIS in
Delhomme JP (1979) Spatial variability and uncertainty in ground- Saveh-Nobaran aquifer, Iran. Environ Earth Sci 71:3827–3843
water flow parameters: a geostatistical approach. Water Resour Tamutou D, Bisso D, Njandjock NP, Nouayou R, Meli’i JL, Teikeu A
Res 15(2):269–280 (2013) Seawater and fresh groundwater demarcation in south-
Djebebe-Ndjiguim CL, Huneau F, Denis A, Foto E, Moloto-a- western Cameroon area. Int J Earth Sci Eng 6(5):985–992
Kenguemba G, Celle-Jeanton H, Garel E, Jaunat J, Mabingui J, Tanriverdi Ç, Alp A, Demirkıran AR, Üçkarde F (2010) Assessment
Le Coustumer P (2013) Characterization of the aquifers of the of surface water quality of the Ceyhan River basin, Turkey.
Bangui urban area, Central African Republic, as an alternative Environ Monit Assess 167:175–184
drinking water supply resource. Hydrol Sci J 58(8):1760–1778 Teikeu AW, Ndougsa-Mbarga T, Njandjock NP, Tabod CT (2012a)
Djeuda Tchapnga HB, Tanawa E, Temgoua E, Siakeu J, Ngo Massana Geoelectric investigation for groundwater exploration in
B (1999) Mode de circulation, mécanismes de recharge et temps Yaoundé area, Cameroon. Int J Geosci 3(3):640–649
relatifs de séjour des eaux des nappes souterraines des altérites Teikeu AW, Njandjock NP, Bisso D, Yene Atangana Q, Sep Nlomgan
du milieu cristallin : cas du bassin versant de l’Anga’a, Yaoundé, JP (2012b) Hydrogeophysical parameters estimation for aquifer
Cameroun. Presse Universitaire de Yaoundé 117–126 characterisation in hard rock environment: a case study from
Kortatsi KB, Tay KC, Anomu G, Hayford E, Dartey A (2008) Yaounde, Cameroon. J Water Res Prot 4(11):944–953
Hydrogeochemical evaluation of groundwater in the lower Offin Uddameri V, Honnungar V, Hernandez AE (2014) Assessment of
basin, Ghana. Environ Geol 53:1651–1662 groundwater water quality in central and southern Gulf Coast
Kuitcha D, Takounjou FA, Ndjama J (2013) Contribution of aquifer, TX using principal component analysis. Environ Earth
hydrochemistry and environmental isotope knowledge of Sci 71:2653–2671
groundwater resources in Yaoundé, Cameroon. J Appl Biosci Uyan M, Cay T (2013) Spatial analyses of groundwater level
67(5):194–5208 differences using geostatistical modeling. Environ Ecol Stat
Kyriakidis P (2004) A geostatistical framework for area-to-point 20:633–646
spatial interpolation. Geogr Anal 36(3):259–289 WHO (World Health Organization) (2004) Guidelines for drinking
Meli’i JL, Bisso D, Njandjock NP, Ndougsa-Mbarga T, Banga FA, water quality, 3rd edn. WHO, Geneva
Manguelé-Dicoum E (2013) Water-table control using ordinary Wu C, Wu J, Luo Y, Zhang H, Teng Y, De Gloria SD (2011) Spatial
Kriging in the southern part of Cameroon. J Appl Sci interpolation of severely skewed data with several peak values
13(3):393–400 by the approach integrating Kriging and triangular irregular
Mvondo H, Ondoa Mvondo, Essono J (2007) Tectonic evolution of network interpolation. Environ Earth Sci 63:1093–1103
the Yaoundé segment of the Neoproterozoic central African Yongsi Nguendo HB (2011) Access and management of drinking
Orogenic Belt in southern Cameroon. Can J Earth Sci water in developing cities: evidence from Yaoundé (Cameroon).
43:433–444 Res J Environ Sci 5(2):124–133
Njandjock NP, Kenfack C, Diab AD, Njeudjang K, Meli’i JL,
Kamseu R (2013) A geostatistical re-interpretation of gravity
surveys in the Yagoua, Cameroon region. Geofis Int
52(4):365–373

123

You might also like