An efficient method for assessing water

International Journal on Soft Computing (IJSC) Vol. 5, No. 2, May 2014
DOI: 10.5121/ijsc.2014.5203 21
AN EFFICIENT METHOD FOR ASSESSING WATER
QUALITY BASED ON BAYESIAN BELIEF NETWORKS
Khalil Shihab1
and Nida Al-Chalabi2
1
College of Engineering & Science, Victoria University, Australia
2
Department of Computer Science, SQU, Oman
ABSTRACT
A new methodology is developed to analyse existing water quality monitoring networks. This methodology
incorporates different aspects of monitoring, including vulnerability/probability assessment, environmental
health risk, the value of information, and redundancy reduction. The work starts with a formulation of a
conceptual framework for groundwater quality monitoring to represent the methodology’s context. This
work presents the development of Bayesian techniques for the assessment of groundwater quality. The
primary aim is to develop a predictive model and a computer system to assess and predict the impact of
pollutants on the water column. The process of the analysis begins by postulating a model in light of all
available knowledge taken from relevant phenomenon. The previous knowledge as represented by the prior
distribution of the model parameters is then combined with the new data through Bayes’ theorem to yield
the current knowledge represented by the posterior distribution of model parameters. This process of
updating information about the unknown model parameters is then repeated in a sequential manner as
more and more new information becomes available.
KEYWORDS
Bayesian Belief Networks, Water Quality Assessment, Data Mining
1. INTRODUCTION
Water is an essential requirement for irrigated agriculture, domestic uses, including drinking,
cooking and sanitation. Declining surface and groundwater quality is regarded as the most serious
and persistent issue and has become as a global issue effecting the people and the ecosystem.
Anthropogenic sources of pollution such as agriculture, industry, and municipal waste, contribute
to the degradation of groundwater quality, which may limit the use of these resources and lead to
health-risk consequences. There are many observable factors contributing to the deterioration of
water quality. These factors need to be monitored and their maximum allowable limits need to be
determined. Decline in water quality is manifested in a number of ways, for example, elevated
nutrient levels, acid from mines, domestic and oil spill, wastes from distilleries and factories, salt
water intrusion and temperature. These factors and others will provide the input data for our
computer system. Efficient water management relies upon efficient monitoring systems that have
the capability to provide information that are decision relevant. Unfortunately, existing
monitoring systems do not always fulfil this objective, where many monitoring systems are
designed to gather data that are redundant and do not add decision-relevant information of value.
Therefore, the needs to acquire data that are decision-relevant, and efficient, establish a need for
the development of cost-effective and flexible analytical methodology for water quality
monitoring networks. Recent attempts based on Artificial Intelligence (AI) were first applied to
the interpretation of biomonitoring data. Other works were based on pattern recognition using
artificial neural networks (NNs). In particular, the supervised learning machines have also been
used in water resources management applications, which have been drawn more attention in the

22
research literature. For instance, the relevance vector machine RVM model has been used in
hydrological applications and groundwater quality modelling showed good results. However,
these methods ignore the dependencies between water quality variables.
Therefore, a Bayesian reasoning approach is employed here. This approach incorporates prior
knowledge about the possible state of a system, and adds new data in a pre-posterior analysis to
produce posterior knowledge of full information about possible system states [1-3]. This approach
enables a comprehensive evaluation of water quality variables and allows establishing public
health concepts. Bayesian methods of statistical inference offer the greatest potential for
groundwater monitoring. This is because these methods can be used to recognize the variability
arising from three different sources of errors, namely, analytical test errors, sampling errors and
time errors, in addition to the variability in the true concentration [4, 5]. The Bayesian methods
can also be used to significantly increase the precision and the accuracy of the test methods used
in a given environmental laboratory [6, 7]. The mobility of salt and other pollutants in steady state
and transient environmental conditions can be predicted by applying Bayesian models to a range
of spatial and temporal scales under varying environmental conditions. Bayesian networks use
statistical techniques that tolerate subjectivity and small data sets. Furthermore, these methods are
simple to apply and have sufficient flexibility to allow reaction to scientific complexity free from
impediment from purely technical limitations [8].
The process of Bayesian analysis begins by postulating a model in light of all available
knowledge taken from relevant phenomenon [9]. The previous knowledge as represented by the
prior distribution of the model parameters is then combined with the new data through Bayes’
theorem to yield the current knowledge (represented by the posterior distribution of model
parameters). This process of updating information about the unknown model parameters is then
repeated in a sequential manner as more and more new information becomes available. In this
work, we studied the Salalah area of Oman because the groundwater has been an important
natural resource and the only available water source other than the seasonal rainfall.
Groundwater quality and pollution are determined and measured by comparing physical,
chemical, biological, microbiological, and radiological quantities and parameters to a set of
standards and criteria. A criterion is basically a scientific quantity upon which a judgment can be
based [10, 11]. In this work, however, we considered only the chemical parameters: total
dissolved solids (TDS), electrical conductivity (EC) and water pH.
2. UNCERTAINTY ANALYSIS]
The Ministry of Water Resources (MWR) maintains data on the concentration of the harmful
substances in the groundwater at Taqah monitoring sites, which are located to the south of the
Sultanate of Oman, in the Salalah plain [12, 13]. We observed that good quality data were
obtained from several monitoring wells in this region. Because of the lack of monitoring wells in
certain areas in that region, we filled in the missing measurements with data obtained from Oman
Mining Company (OMCO) and Ministry of Environmental and Regional Municipalities (MRME)
[14].
Data for water quality assessment are normally collected from various monitoring wells and then
analyzed in environmental laboratories in order to measure the concentration of a number of
water quality constituents. We realized that the methods used by these laboratories do not
emphasize accuracy. There is a lack of awareness among both laboratory and validation personnel
regarding the possibility of false positives in environmental data. In order to overcome this
problem and to have representative data, we, therefore, used the following modified Bayesian
model to that developed by Banerjee, Planting and Ramirez [10], to preprocessing the datasets
used for the development of the Bayesian Networks.

23
2.1. Bayesian Models
The formulation of the model is as follows:
Let S denote a particular hazardous constituent of interest. Since the concentration of the
substance may vary from well to another, it is necessary to consider each well separately. Let xt=
(xt1, xt2, xt3, xtm) be the vector of m measurements of the concentration of S in m distinct water
samples from a given well at a given sampling occasion where (m>=1) and (t=1, 2, . . .). Each
measurement consists of the true concentration of S plus an error.
Let Xt be the true concentration of S in the groundwater at sampling occasion t. If we assume that
the true concentration Xt is unknown and is a random variable, the model evaluates the posterior
distribution of Xt given the sample measurements xt at sampling occasion t.
Using the normality assumption and given Xt = xt and δ2
, the concentration measurements in xt
represent a random sample of size m for random distribution with mean xt and variance δ2
.
We assume that the parameters xt and δ2
of the normal distribution are random variables with
certain prior probability distribution. Therefore, the model for prior distribution of Xt and δ2
can
be presented as follows:
For t =1, 2… and given δ2
the conditional distribution of Xt at sampling occasion t is a normal
distribution with mean μt-1 and variance δ2
t-1 δ2
. The marginal distribution of δ2
is an inverted
gamma distribution with parameter βt-1 and νt-1.
This model uses the following prior distribution, which represents the concentration
measurements before the first sampling.
The pdf of the prior distribution of X0 is:
2)12(2
0
0
0
00
0
00
0
2
1
1)(
+−


























−
+=
v
v
x
v
xf


(2.1)
which is the pdf of the student’s t-distribution with 2v0 degrees of freedom, location parameters
μ0 and variance δ0
2
β0/ν0.
Now suppose that the observations are available on the concentration of S, given the sample Xt
the posterior marginal distribution of Xt is a student’s t-distribution with 2vt degree of freedom,
location parameters μt and variance δt
βt/νt where the pdf has the form:
2)12(2
2
1
1)/(
+−















 −
+=
tv
ttt
tt
t
tt
x
v
xxf


(2.2)
where:
[ ])1(2/)(2/)( 2
11
1
1 −−
=
− +−+−+= ∑ ttt
m
j
tjtt mxmxx 
2/1 mvv tt += −

24
2
1
2
11 1/()( −−− ++= ttttt mxm  ) (2.3)
)1/( 2
1
2
1
2
−− += ttt m
mxx
m
j
tjt /
1
∑=
=
It is obvious from the equation of μt the sequential nature of this posterior distribution. Therefore,
in order to present the true unknown concentration of the substance S in the well under
consideration, it is frequently more convenient to put a range (or interval) which contains most of
the posterior probability. Such intervals are called highest posterior density (HPD) intervals. Thus
for a given probability content of (1-α), 0< α<1, a 100(1- α) percent HPD interval for Xt, is given
by:
tttvt t
t  )2/(2±
(2.4)
when t2vt(α/2) is the 100(1- α/2) percentile of the student’s t-distribution with 2vt degree of
freedom.
2.2. Bayesian Algorithm
In brief, the monitoring algorithm, which is based on the Bayesian model, is as follows:
(1) Fix a value of α (0< α <1) based on the desired confidence level. In this case, we chose α
to be 0.01.
(2) Since we do not have enough data to work with, we used the same parameters of the prior
distribution used in the model of Banerjee, Plantinga and Ramirez. These parameters are :
β0= 0.0073 , ν0=2.336 , μ0= 9.53 , δ0
2
=3056.34
(3) At each sampling occasion t , ( t= 1,2,...), compute the parameters βt , νt , μt and δt of the
posterior distribution Xt given the set of observations in xt on the concentration of S
available from a given well in a given site using (2.3). Compute LHPD and UHPD using
these parameter estimates and (2.4).
(4) Plot μt, LHPD, and UHPD that are obtained in step 3 above against sampling occasion t.
(5) For the next sampling occasion, update the values of the parameters βt, νt, μt and δt using
(2.3) and the datasets just obtained. Recomputed LHPD, and UHPD using the updated
parameter values in (2.4) and repeat step 4 above.
Some of these datasets needed to be scaled down using the following normalization technique:

−
=
x
x
, where
nxx
n
i
i∑=
, and 1
22
−
−
=
∑
n
xnx
n
i
i

2.3. Implementation
The pre-processing system is implemented on PC platform using Visual Basic programming
language.
Table 1 presents the concentration data for TDS (Total Dissolved Solids) for Well 001/577 in the
Taqah area. In particular, the table shows the true concentration data for TDS produced by our
pre-processing system.

25
Table 1. Concentration Data of TDS for Well001/577 in the Salalah plain, where OC stands for Observed
Concentration and ETC stands for Expected True Concentration.
Te OC LHPD ETC UHPD
84 1.147 0.85 1.15 1.45
85 1.106 1 1.13 1.26
86 1.938 1.12 1.4 1.68
87 2.237 1.33 1.61 1.88
88 3.857 1.6 2.06 2.52
89 3.834 1.91 2.35 2.79
90 3.957 2.18 2.58 2.98
91 3.761 2.38 2.73 3.08
92 4.3 2.58 2.9 3.23
93 3.958 2.72 3.01 3.3
94 1 2.54 2.83 3.11
95 3.714 2.64 2.9 3.16
96 3.65 2.73 2.96 3.19
97 3.381 2.78 2.99 3.2
98 3.396 2.83 3.02 3.2
99 3.477 2.87 3.04 3.22
00 3.498 2.91 3.07 3.23
01 3.23 2.93 3.08 3.23
02 3.243 2.95 3.09 3.22
03 3.267 2.97 3.1 3.22
04 3.297 2.99 3.11 3.22
3. BAYESIAN NETWORKS
After the pre-processing stage, we constructed a Bayesian Network (BN) by using the Hugin
system. We then used this BN as an initial building network for the construction of two Dynamic
Bayesian Networks in order to predict the impact of pollution on groundwater quality [15, 16].
3.1. Dynamic Bayesian Networks (DBNs)
DBNs extend Bayesian Networks from static domains to dynamic domains [17, 18]. This is
achieved by introducing relevant temporal dependencies between the representations of the static
network at different times.
The main characteristic of DBNs is as follows:
Let Xt be the state of the system at time t, and assume that
(1) The process is Markovian, i.e.,
P(Xt/X0, X1, . . ., Xt-1)= P(Xt/Xt-1)
(2) The process is stationary or time-invariant, i.e.,
P(Xt/Xt-1) is the same for every t.

26
Therefore, we just need P(X0), which is a static Bayesian network (BN), and P(Xt/Xt-1), which is a
network fragment, where the variables in Xt-1 have no parents, in order to have a Dynamic
Bayesian Network (DBN).
3.2. Bayesian Networks Development
Among more than twenty wells in the Taqah area, we selected only four wells for this study.
Those four wells have had, to the greatest extent, complete data measurements and provide
sufficient information for the assessment of the groundwater quality for this area.
The electrical conductivity (EC) of the water has been used as a measure for the salinity hazard of
the groundwater used for irrigation in the Salalah plain. The total dissolved solid (TDS) limit is
600 mg/L, which is the objective of the current plan of the MWR. TDS contains several dissolved
solids but 90% of its concentration is made up of six constituents. These are: sodium Na,
magnesium Mg, calcium Ca, chloride Cl, bicarbonate HCO3 and sulfate SO4. We, therefore,
considered only these elements in the calculation of TDS.
We also used the following relationship between TDS and EC.
TDS = A * EC; where A is a constant with value between 0.65 and 0.77.
Both TDS and EC can affect water acidity or water pH. Solute chemical constituents are variable
in high concentration at lower pH (higher acidity). On the other hand, acidity allows migration of
hydrogen ions (H+), which is an indication of conductivity. Therefore, our work concentrated on
the following relations.
TDS  EC, EC  pH, TDS  pH
Reaching to these relations we used two learning approaches to construct and parameterize a
simple static BN that have three nodes, each node represents a groundwater quality constituent
(TDS, EC or pH). Learning basically consists of two different components: 1) learning the
network structure, 2) learning the conditional probability distributions.
For the first component, we used the Hugin system that supports structure and parameter learning
in Bayesian networks. We also developed a program written in C++ to generate the conditional
probabilities for TDS, EC and pH using Table 2 as input.
Once the static BN model (static model) for each monitoring well was built, parameterized and
tested, we used these models as initial building networks in the construction of OOBNs. Figure 1
models the time slices for each well characterizing the temporal nature of identical model
structures, where the initial building network, see Figure 2, describes a generic time-sliced
network.

27
Table 2. TDS, EC, and pH data for the well Well 001/577.
Yr TDS
mg/L
EC
µS/cm
pH
84 542.7 548 7.85
85 525.5 548 7.8
86 565.4 579 7.75
87 604.2 588 7.57
88 541.8 601 7.43
89 565.9 625 7.34
90 558.6 638 7.32
91 640.4 798 7.27
92 754.5 739 7.24
93 798.7 758 7.28
94 746.4 799 7.29
95 615.8 514 7.3
96 737.5 619 7.28
97 753.6 869 7.19
98 935.6 558 7.15
99 1174 855 7.15
0 1021 796 7.06
1 1067 855 6.98
2 1223 844 6.94
3 1055 881 6.9
Figure 1. The OOBN representing three time-sliced networks

28
Figure 2. The initial building block representing one time-sliced network
Since our Bayesian Networks are tractable models, we also implemented the exact inference for
the network described in Figure 8 and compared the results with that produced by OOBN. Figures
3 and 4 show the KL-divergence between the true and the approximate distribution [10]. Since
the KL-distance converges to zero, this is an indication of the accuracy and reliability of OOBN
Figure 3. True and approximate probability distributions.
Figure 4. The KL-divergence between the true and the
estimate distributions over all variables.
4. USING CLASSICAL TIME SERIES FOR THE ASSESSMENT OF
GROUNDWATER QUALITY
The purpose of this section is to apply the classical time series analysis to groundwater quality
data and to compare the results with that obtained by the application of Dynamic Bayesian
Networks (DBNs). The continuous and regular monitoring data of electrical conductivity (EC),
total dissolved solid (TDS), pH measured by the Ministry of Water Resources (MWR) were also
used here for the time series analysis.
Time series analyses of water supply wells with respect to the concentration of chemical
constituents are presented in Figures 5-10.

29
Total dissolved solids (TDS) are a measure of the dissolved minerals in water and also a measure
of drinking water quality. There is a secondary drinking water standard of 500 milligrams per liter
(mg/L) TDS; water exceeding this level tastes salty. Groundwater with TDS levels greater than
1500 mg/L is considered too saline to be a good source of drinking water. Figure 5 shows the
concentration of TDS for the well Well001/577 for a period of twenty one years.
The fluctuation of the concentration of the chloride (Cl), sodium (Na), and calcium (Ca) with
respect to time is shown in Figure 7. The values were averaged during the initial analysis as there
were no significant differences among the monthly data. Chloride values above 250 mg/l give a
slight salty taste to water which is objectionable by many people.
Relationships between TDS, EC and pH are examined using multiple regression analysis, see
Figure 7. Multiple regression analysis is used to explain as much variation observed in the
response variable as possible, while minimizing unexplained variation from “noise”. The results
of this analysis are used to produce the moving average chart, Figure 9, and the linear regression
chart, Figure 10. We used Excel Business Tools [19], Microsoft Excel, and Matlab for producing
these and other charts.
y = 3 3.9 2 8x + 4 1 2.5 3
R 2
= 0 .8 2 2 3
0
2 0 0
4 0 0
6 0 0
8 0 0
1 0 0 0
1 2 0 0
1 4 0 0
1 9 8 4
1 9 8 6
1 9 8 8
1 9 9 0
1 9 9 2
1 9 9 4
1 9 9 6
1 9 9 8
2 0 0 0
2 0 0 2
2 0 0 4
mg/l
T D S
L in e a r
( T D S )
Figure 5. Fluctuation of TDS concentration for the well Well001/577
y = 0 .0 1 2x - 2 1.1 0 5
R 2
= 0 .5 1 8 3
2 .5 5
2 .6 0
2 .6 5
2 .7 0
2 .7 5
2 .8 0
2 .8 5
2 .9 0
2 .9 5
1 9 8 4 1 9 8 9 1 9 9 4 1 9 9 9 2 0 0 4 2 0 0 9
EC
Figure 6. EC concentration is poorly represented for the well Well001/577

30
0
50
100
150
200
250
300
350
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
mg/l
M g
So4
N a
C a
K
C l
Figure 7. Fluctuation of the concentration of the major chemical constituents for Well001/577 for a period
of 21 years
Figure 8. Excel templates for financial analysis and business productivity
from Excel Business Tools
As is shown in Figure 7 that the trend is as follows:
TrendWQ=19.01*TDS - 5.42*EC -270.16*pH + 205.14
5 . 0 0
0 . 0 0
5 . 0 0
1 0 . 0 0
1 5 . 0 0
2 0 . 0 0
2 5 . 0 0
1 9 8 0 1 9 8 5 1 9 9 0 1 9 9 5 2 0 0 0 2 0 0 5
Y e a r
Trend
Figure 9. Moving average chart of 2-year period for groundwater quality trend
Figure 10. A curve fitting chart showing groundwater quality trend over time

31
Figure 9 shows the groundwater quality trend over time (linear regression). The trend has the
following properties:
Linear model Poly1:
f(x) = p1*x + p2
Coefficients (with 95% confidence bounds):
p1 = 0.8954 (0.7962, 0.9947)
p2 = 1.332 (0.08589, 2.579)
Goodness of fit:
SSE: 32.91
R-square: 0.9494
Adjusted R-square: 0.9467
RMSE: 1.316
Although the classical time series models are used here to assess the presence and strength of
temporal patterns of groundwater quality. These models are based on the assumption of stationary
(i.e. time invariant). They have been widely used in many domains such as financial data and
weather forecasting. Yet these models do not readily adapt to domains with dynamically changing
model characteristics, as is the case with groundwater quality assessment. In addition to the above
mentioned assumption, the classical models are restricted in their ability to represent the general
probabilistic dependency among the domain variables and they fail to incorporate prior
knowledge.
The observed groundwater quality data are irregularly spaced and not predetermined as in the
case with ordinary time series. This may cause the traditional time series techniques to be
ineffective (Prediction: what is the predicted value for one period a head). It is evident that the
time series casts doubts on the positive or negative effects of any chemical constituent on the
groundwater quality for the long run, and is thus not as clear and reliable as in the case of using
Dynamic Bayesian Techniques. While some groundwater quality constituents, such as chloride
and TDS, show an increasing trend, the other constituents, such as pH, Mg, and SO4 do not
demonstrate obvious trends. Therefore, we can draw a reliable conclusion on the cause of the
increasing trend of the groundwater quality and we cannot investigate the effect of the increasing
or decreasing other constituents, such as pH and EC. In addition to this ignorance of the cause-
effect relationships, classical time series models assume the linearity in the relationships among
variables and normality of their probability distributions.
5. CONCLUSION AND FURTHER WORK
This work presents the assessment of groundwater quality. Bayesian methods have been
investigated and shown to offer considerable potential for use in groundwater quality prediction.
These methods are based on reasoning under conditions of uncertainty. This work is the first step
towards having a comprehensive network that contains the other variables that are considered by
the researchers significant for the assessment of groundwater quality in the Salalah plain in
particular.
Also we showed that the classical time series models do not readily adapt to domains with
dynamically changing model characteristics, as is the case with groundwater quality assessment.
This is mainly because these models are restricted in their ability to represent the general

32
probabilistic dependency among the domain variables and they fail to incorporate prior
knowledge.
REFERENCES
[1] Borsuk, M.E., C.A. Stow, and K.H. Reckhow, A Bayesian network of eutrophication models for
synthesis, prediction, and uncertainty analysis. Ecological Modelling, 2004. 173(2): p. 219-239.
[2] Brandherm, B. and A. Jameson, An extension of the differential approach for Bayesian network
inference to dynamic Bayesian networks. International Journal of Intelligent Systems, 2004. 19(8): p.
727-748.
[3] Zou, M. and S.D. Conzen, A new dynamic Bayesian network (DBN) approach for identifying gene
regulatory networks from time course microarray data. Bioinformatics, 2005. 21(1): p. 71-79.
[4] Kim, S., S. Imoto, and S. Miyano, Dynamic Bayesian network and nonparametric regression for
nonlinear modeling of gene networks from time series gene expression data. Biosystems, 2004. 75(1):
p. 57-65.
[5] Shihab, K., Analysis of Water Chemical Contaminants: A Comparative Study Applied Artificial
Intelligence (AAI), Vol 22, No. 4, pp. 352-376, 2008.
[6] Lung, W.-S., Water quality modeling: application to estuaries. Vol. 3. 1993: CRC press.
[7] Varis, O., Belief networks for modelling and assessment of environmental change. Environmetrics,
1995. 6(5): p. 439-444.
[8] Stow, C.A., et al., Comparison of estuarine water quality models for total maximum daily load
development in Neuse River Estuary. Journal of Water Resources Planning and Management, 2003.
129(4): p. 307-314.
[9] Russell, S.J., et al., Artificial intelligence: a modern approach. Vol. 2. 1995: Prentice hall Englewood
Cliffs.
[10] Banerjee, A.K., Monitoring groundwater quality. 1985, Department of Statistics: University of
Wisconsin.
[11] Shihab, K. and N. Chalabi, Dynamic Modeling of Ground‐Water Quality Using Bayesian
Techniques1. JAWRA Journal of the American Water Resources Association, 2007. 43(3): p. 664-
674.
[12] Dames and Moore, Investigation of The Quality of Groundwater Abstracted from the Salalah Plain:
Dhofar Municipality. 1992.
[13] Law on the Protection of Water Resources, promulgated by Decree of the Sultan No. 29 of 2004, and
its implementing regulations (Regulations for the organization of wells and aflaj, and Regulations for
the use of water desalination units on wells), (in Arabic). 2004, Ministry of Water Resources (MWR).
[14] Shihab, K, Al-Chalabi N, Treatments of Water Quality Using Bayesian Reasoning. Lecture Notes in
Computer Science, 2004. 3029: p. 728–738.
[15] Nielsen, T.D. and Jensen, F.V., Bayesian networks and decision graphs. 2009: Springer.
[16] Korb, K.B. and A.E. Nicholson, Bayesian artificial intelligence. 2003: cRc Press.
[17] HUGIN Expert Brochure. HUGIN Expert 2014; Available from: http://www.hugin.com.
[18] Kjærulff, U., dHugin: A computational system for dynamic time-sliced Bayesian networks.
International journal of forecasting, 1995. 11(1): p. 89-111.
[19] Excel Business Tools. 2014 [cited 2014 10/01/14]; Available from:
http://www.excelbusinesstools.com/index.htm.

An efficient method for assessing water

More Related Content

What's hot (18)

Viewers also liked (17)

Similar to An efficient method for assessing water (20)

Recently uploaded (20)

An efficient method for assessing water