See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/256375672

Assessing groundwater vulnerability to contamination using geostatistical

approach

Conference Paper · September 2011

CITATIONS READS

0 282

3 authors, including:

Roha Jamil Mohd Nadzri Md Reba

University of Malaysia, Kelantan Universiti Teknologi Malaysia
3 PUBLICATIONS   4 CITATIONS    72 PUBLICATIONS   254 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Active Water Leak Detection Utilizing Integrated Non-Destructive Detection Techniques For Non-Revenue Water (NRW) Reduction View project

Weather radar processing for precipitation estimation View project

All content following this page was uploaded by Roha Jamil on 01 June 2014.

The user has requested enhancement of the downloaded file.

 ISG & ISPRS 2011, Sept. 27-29, 2011 – Shah Alam, MALAYSIA

Assessing groundwater vulnerability to contamination

using geostatistical approach
Rohazaini Muhammad Jamil1, Mohamad Nor Said2 and Mohd Nadzri Md Reba3

Department of Geoinformatics, Faculty of Geoinformation and Real Estate,

Universiti Teknologi Malaysia, 81310 UTM Skudai, Johor, Malaysia.
1
[email protected], [email protected], [email protected]

Abstract

Due to a rapid development of industrial and population growth in Malaysia, the demand
for water has increased tremendously. This has caused the shortage of water supply and the
new water sources have to be identified – groundwater is one of the alternative. In the north
eastern state of Kelantan, most people rely on the groundwater as their water resource.
However, due to uncontrolled development and human activities, groundwater is subjected
to pollution and geochemical contamination. This paper is intended to assess groundwater
vulnerability to contamination using geostatistical approach instead of using the typical
DRASTIC (Depth, Recharge, Aquifer, Soil, Topography, Influence of the vadose zone and
Conductivity) model. Geostatistical approach has the capability to predict a specific
vulnerability of contamination risk as an alternative to the DRASTIC model which is
unable to provide the measurable information of geochemical characteristics. The results of
the modelling are produced and presented in the form of Prediction Map and Probability
Map. The finding of the study can be used to identify the geochemical concentration based
on Department of Environment (DOE) standards.

Keywords: Groundwater vulnerability to contamination, geostatistics, DRASTIC model

1. Introduction

The concept of vulnerability of groundwater to contamination was introduced by Margat (1968) and
the term “vulnerability” was defined as the degree of protection that the natural environment
provides against the spread of pollution in groundwater. However, this definition has gradually
changed, e.g. Hrkal (2001) describes groundwater vulnerability as the tendency and likelihood for
general contaminants to reach the water table after introduction at the ground surface.

Groundwater is vulnerable to contamination by anthropological activities and it is very difficult to

remediate once contaminated. To properly manage and protect the resource, it is therefore important
to determine areas with more aspects of vulnerable to contamination. “Vulnerability” is the degree
of which human or environmental systems are likely to experience harm due to perturbation or
stress, and can be identified for a specific system, hazard, or group of hazards (cited by Liggett and
Talwar, 2009 in Popescu et al. 2008). Assessment of the vulnerability often results in a map of areas
where the resource is vulnerable to contamination due to the surface activities. The assessment puts
high priority to the contaminated areas for further investigation, protection, and monitoring.

Groundwater vulnerability assessments are meant to synthesizing complex hydrogeologic

information into a feasible map to be used by planners, decision and policy makers, geoscientists,
and the public. Development of the vulnerability maps is useful for many aspects of water

1
 ISG & ISPRS 2011, Sept. 27-29, 2011 – Shah Alam, MALAYSIA

management, including prioritizing areas for monitoring, protection, and further investigation; and
the development of risk assessments, resource characterization, and education. In general,
vulnerability assessments are categorized as (i) index (and overlay) methods; (ii) statistical
methods; or (iii) process methods (Focazio et al. 2002). The examples of models for each of these
methods are shown in Table 1.

Table 1: Examples of each of the three methods of vulnerability assessment (Liggett and Talwar, 2009)

Methods Models (examples)

Index and Overlay methods DRASTIC
Statistical Methods Logistic Regression
Process Methods Numerical Models

Index and overlay method is the most popular one that has been applied in many studies for
assessing groundwater vulnerability. The most eminent model (which is used in this study) is called
DRASTIC model (Liggett and Talwar, 2009). Another practice of assessing vulnerability is by using
statistical method. This method is based on observations (i.e. not on expert opinion) and
anthropogenic factors that are easy to update as new available information. In statistical approach
one of the typical methods is the logistic regression that describes the relationship between a
categorical outcome (response variable) and a set of covariates (predictor variables). Apart from
this, the process based simulation model approach - process based method introduced by Focazio et
al. (2002) - is also being used. Here additional interpretation by the water resource decision maker
and scientists’ consultation is required in order to meet certain policy and management objectives.
Basically the numerical model is applied in this process based method.

As compared to these two methods, the DRASTIC model (by means of category of overlay and
index method) is prominently used in identifying areas at which groundwater supplies are most
susceptible to contamination (Aller et al., 1987 and Sener et al., 2009). This method is the typical
groundwater vulnerability assessment for its advantageous of relatively inexpensive,
straightforward, and relying on data that are commonly available or estimated, and producing end-
products that are easy to interpret and incorporate into the decision-making process (Focazio et al.,
2002).

Though, the DRASTIC model has yet to be analysed in order to find the potential improvement as
its limitation has been revealed by the past studies. Evans and Maidment (1995) claimed that the
DRASTIC model is lack of methodological foundation in which introduction of contaminant
loading in the DRASTIC model unfortunately contributes very little to the correlation of the actual
contamination occurrence with the derived scores.

Despite of these limitations, a study conducted at the Centre of Research in Water Resources, Texas
has applied statistical methods in mapping vulnerability groundwater to nitrate contamination
(Evans and Maidment, 1995). The study claimed that a statistical approach was chosen for two
reasons. The first is the dissatisfaction with index or overlay methods and process based models.
The second is the appropriateness of this approach to GIS-based analysis. This method is the least
common vulnerability assessment in the literature and it is normally used as a test for other
methods. Besides it is straightforward to apply as an index or overlay method, but with a more
defensible foundation.

2
 ISG & ISPRS 2011, Sept. 27-29, 2011 – Shah Alam, MALAYSIA

Geostatistical methods such as kriging are a ubiquitous approach that quantitatively estimates the
distribution of water quality parameters. However there is no work related to the application of
geostatistical method in the groundwater vulnerability assessment has neither been published nor
verified (GAO, 1992). Therefore, the main objective of this study is to assess groundwater to
contamination by using the geostatistical method to model the continuous surface of geochemical
concentration and finally to determine which spot (i.e. well) has exceeded the threshold value of
geochemical concentration. This paper discusses the typical issues in DRASTIC model and
introduces geostatistics as an alternative approach.

2. Groundwater vulnerability assessment based on DRASTIC model

DRASTIC has been used for mapping aquifer vulnerability in porous aquifers (Aller et al. 1987).
The DRASTIC method is derived from ratings and weights associated with the seven parameters
namely the depth to groundwater (D), the net recharge (R), the aquifer media (A), the soil media
(S), topography (T), influence of the vadose zone (I) and the hydraulic conductivity (C). Each
parameter is subdivided into ranges with different ratings assigned in a scale of 1 to 10 (least to
highest contamination potential). A higher DRASTIC index shows greater groundwater pollution
vulnerability (Sener et al., 2009).

Yet, due to its limitation, some studies have applied some modifications and improvements to the
DRASTIC procedures (Thirumalaivasan et al., 2003; Dixon, 2005; Panagopoulus et al., 2006).
According to Rosen (1994), some important scientific parameters such as sorption capacity, travel
time and dillution are not taken into account directly. Besides, it is very difficult to test the accuracy
of the model because it requires properties of pollutant to be assumed as a model and deposited for
all over the test area at a uniform concentration and for a considerable period of years in order to
allow the hydrogeological setting to respond.

Due to the fact that DRASTIC approach is unable to provide a measurable information of chemical
characteristics (which is very important to the environmentalists, scientists or hydrogeologists), thus
there is a need to introduce an alternative method to cope with these drawbacks.

2.1 Geostatistics as an alternative approach

The information of chemical characteristics is very important to the environmentalists, scientists or

hydrogeologists to quantitatively define the characteristics of surface and volumes of regionalized
variables which have been sampled only at discrete points. Geostatistical approach has been
identified to be an alternative way in assessing groundwater vulnerability to contamination. The
geostatistics features a statistical model to appropriately estimate the continuous area (i.e.
groundwater contamination phenomenon) in a discrete form.

The principles of geostatistics have been applied to a variety of areas in geology and other scientific
disciplines (Nelson et al., 1999, Hamad, 2009). The major concern is that of defining quantitatively
the characteristics of surface and volumes of regionalised variables which have been sampled only
at discrete points. This has raised a question on how possible one to estimate the value of a variable
in a site if the scientists and engineers are provided with a set of sampled data collected at specific
location within an area under investigation.

3
 ISG & ISPRS 2011, Sept. 27-29, 2011 – Shah Alam, MALAYSIA

Therefore, Geographic Information Systems (GIS) coupled with geostatistics application have the
capability to help in virtually display and present the assessment results. The geostatistics is a handy
tool that works with continuous data such as rainfall, temperature, geology, or soils to create a
surface (probability and prediction surfaces). Here, one of the sought-after features is kriging that
applies interpolation and prediction at the unsampled location and utilises spatial relationships to
give a better and accurate predictions. The kriging interpolator is considered the most sophisticated
and accurate way to determine the intensity of a phenomenon at unmeasured locations. Kriging
weights surrounding measured values are based not only on the distance between measured points
and the prediction location but also on the overall spatial arrangement of the measured points.

3. Methodology

There are two major phases of methodology in this study, (1) data collection and (2) modelling
using geostatistical method. Data collection phase is illustrated in Figure 1.

Figure 1: Data collection phase.

The study area is located in the north of Kelantan which is the most populated area with
groundwater wells mostly found. It is a potential spot in which most of the groundwater sampling
and monitoring activities are systematically carried out by the Department of Mineral and
Geoscience (JMG). Five geochemical parameters have been selected to be modelled as it shows a
significant fluctuation of concentration level from year 2004 till year 2007 (Rohazaini, 2008)

The second phase is data modelling using geostatistical approach as shown in Figure 2. In this
study, Geostatistical Analyst of ArcGIS 9.2 is used to model the groundwater contamination maps.
Instead of providing various interpolation techniques, the geostatistical feature provides (prior to
mapping) Exploratory Spatial Data Analysis (ESDA) tools that help to assess the statistical
properties of the data. In the exploratory spatial data analysis, variety of output map types
(prediction, error of prediction, probability and quantile) are created using kriging and co-kriging
variants (ordinary, simple, universal, indicator, probability and disjunctive) and auxillary tools (data
transformation, declustering, and detrending).

Two types of kriging are carried out namely Ordinary Kriging and Indicator Kriging. The Ordinary
Kriging produces the prediction map of contamination concentration while the Indicator Kriging is
more suitable to produce the probability one. The purpose of creating the prediction map is to
model the continuous surface of contamination concentration based on ground measurement (e.g.

4
 ISG & ISPRS 2011, Sept. 27-29, 2011 – Shah Alam, MALAYSIA

concentration value of the manganese, iron, lead, nitrate and chloride) taken from 2004 to 2007. On
the other hand, the probability map is used to locate and identify the risky area and well which
exceeds the critical threshold/standards of the Department of Environment (DOE). Finally, these
two maps were overlaid together to produce groundwater contamination risk map.

Figure 2: Modelling phase

4. Results and discussion

The groundwater contamination risk map shows the geostatistics is a proven variant that gives good
estimate of probability of the area in the North Kelantan contaminated by different geochemical
elements in 2004 to 2006. The map has been modelled in five types of geochemical elements (iron,
chloride, nitrate, manganese and lead). As an example, Figure 3 shows the probability map of
nitrate exceeding the DOE standard in 2006. The excessive level of contamination is presented in
dark purple contour and being determined as risky area that is vulnerable to contamination.

In the decision-making process, care must be taken in using a map of predicted contaminated wells
for identifying unsafe areas because it is necessary to understand the uncertainty of the predictions.
For example, suppose the critical threshold of DOE standard value is 10 mg/litre for nitrate and any
locations exceed this value will be determined. As Geostatistical Analyst provides a number of
methods that can perform this task, indicator kriging have been chosen as it is the simplest available
model to be applied. This technique does not require the dataset to conform to a particular
distribution. The data values are transformed to a series of 0s and 1s according to whether the
values of the data are below or above a threshold. If a threshold of 10 mg/litre is used, any value
below this threshold will be assigned a value of 0, whereas the values above the threshold will be
assigned a value of 1. Indicator kriging then uses a semivariogram model that is calculated from the
0–1 dataset.

5
 ISG & ISPRS 2011, Sept. 27-29, 2011 – Shah Alam, MALAYSIA

For this study, the maps have been modelled based on five types of geochemical elements (iron,
chloride, nitrate, manganese and lead). Figure 4 shows one of the results from this study in
modelling and determining the percentage of nitrate concentration exceed the DOE standard. For
example, in year 2006, six monitoring well locations have been affected by nitrate contamination
and having a 12.6% possibility of exceeding the DOE standards as the calculation value for the dark
purple contours are 0.126821. The exceedence of specified threshold (DOE standards) can be
considered as that particular area is risky and vulnerable to contamination. This analysis is very
important to the regulatory bodies or Department of Environment to make decisions in
monitoring water quality in the study area.

Figure 3: The dark purple contour shows the probability of

nitrate concentration exceeds the DOE standards in 2006

Figure 4: The resultant map shows the information of which

location/well that is having a 12.6% possibility of
exceeding DOE standards

6
 ISG & ISPRS 2011, Sept. 27-29, 2011 – Shah Alam, MALAYSIA

5. Conclusion
Spatially related statistics or geostatistics can be applied in hydrogeology field especially in
groundwater vulnerability to contamination issues. This study is useful in developing groundwater
protection strategies and for optimal monitoring, water management decision makers, as well as
regulatory bodies’ attempts for the regions in which the prevention of groundwater contamination
seems most needed. For example, it will help planners and decision makers on proper land use and
water resource management such as to avoid high risk areas when locating a site of pollution
potential or selecting the areas for waste disposal and industrial sites. In future, geostatistics is
expected be applied spatially in natural resources evaluation for practitioners in such diverse fields
as soil science, mining, petroleum, remote sensing, hydrogeology, and the environmental sciences.

Acknowledgements
The authors are thankful to the Mineral and Geoscience Department of Kota Bharu Kelantan for
their help and priceless support in ensuring the success of this study.

References
Aller, L., Bennett., Lehr, J.H., Petty, R.J., Hackett, G., (1987), DRASTIC: A Standardised System
For Evaluating Ground Water Pollution Potential Using Hydrogeologic Settings, US
Environmental Protection Agency Report (EPA/600/2-87/035), Robert S. Kerr
Environmental Research Laboratory, 455 pp.

Dixon, B., (2005), Groundwater Vulnerability Mapping: A GIS And Fuzzy Rule Based Integrated
Tool, Journal Of Applied Geography 25(2005) 237-347.

Evans, T.A, and Maidment, D.R., (1995), A Spatial And Statistical Assessment Of The
Vulnerability Of Texas Groundwater To Nitrate Contamination, Centre For Research In
Water Resources, The University of Texas, Austin. Retrieved from
http://civil.ce.utexas.edu/centers/crwr/reports/online.html (Last accessed 27 January 2011).

Focazio, M.J., T.E. Reilly, M.G. Rupert, and D.R. Helsel., (2002), Assessing Groundwater
Vulnerability To Contamination: Providing Scientifically Defensible Information For
Decision Makers, US Department of Interior and US Geological Survey, Reston, VA. US
Geological Survey Circular No. 1224.

GAO (General Accounting Office) (1992). Goundwater Protection: Validity and Feasibility of
EPA's Differential Protection Strategy. (GAO/PEMD-93-6) Washington, D.C.: General
Accounting Office. Retrieved from http://www.legistorm.com/showFile/L2xzX3Njb3JlL
2dhby9wZGYvMTk5Mi8xMg==/ful22507.pdf (Last accessed 26 January 2011).

Hamad, S., (2009), Geostatistical Analysis Of Groundwater Levels In The South Al Jabar Al
Akhdar Area using GIS, Proceedings Of GIS Ostrava 2009, January, 25-28. Retrieved from
http:www.gis.usb.cz/GIS_Ostrava/GIS_Ova_2009/Sbornik/Lists/Papers/005.pdf (Last
accessed 5 February 2011).

7
 ISG & ISPRS 2011, Sept. 27-29, 2011 – Shah Alam, MALAYSIA

Hrkal, Z., (2001), Vulnerability Of Groundwater To Acid Deposition, Jiserke Mountains, Northern
Czech Republic: Construction And Reliability Of A GIS-Based Vulnerability Map, Journal
Of Hydrogeology, 9(2001), 348-357.

Liggett J.E., and Talwar, S., (2009), Groundwater Vulnerability Assessments And Integrated
Water Resource Management, Streamline Watershed Management, Vol. 13/no.1 Bulletin
Fall 2009, article 4, 18-29 .

Margat, J., (1968), Groundwater Vulnerability Maps, Conception Estimation Mapping (in French),
EEC Institute European de I’Eau, Paris, France, p11.

Nelson, M.R., Orum, T.V., Nadeem, A., and Garcia, R.J., (1999), Applications Of Geographic
Information Systems And Geostatistics In Plant Disease Epidemiology And Management,
The American Phytopathological Society, Plant Disease, 4(83), 308-319.

Panagopoulus, G.P., Antonakos, A.K., and Lambrakis N.J., (2006), Optimizing Of The DRASTIC
Method For Groundwater Vulnerability Assessment Via The Use Of Simple Statistical
Methods And GIS, Hydrogeology Journal 14(2006), 894-911.

Popescu, I.C., N. Gardin, S. Brouyére, and A. Dassargues., (2008), Groundwater Vulnerability

Assessment Using Physically-Based Modelling: From Challenges To Pragmatic Solutions,
in ModelCARE 2007 Proceedings, Calibration and Reliability in Groundwater Modelling,
J.C. Refsgaard, K. Kovar, E. Haarder, and E. Nygaard (editors), Denmark, IAHS
Publication No. 320.

Rohazaini, M.J., (2008), Permodelan Pencemaran Geokimia Air Tanah Dalam Lapisan Aluvium Di
Negeri Kelantan Menggunakan Geostatistical Analysts (Arcgis 9.2), Bachelor of Science
Geoinformatics Thesis, Faculty of Engineering and Science Geoinformation, Universiti
Teknologi Malaysia.

Rosen, L., (1994), A Study Of The DRASTIC Methodology With Emphasis On Swedish
Conditions, Ground Water 32(1994), 278–285.

Sahoo, R.N., (2003), Geostatistics In Geoinformatics For Managing Spatial Variability, Indian
Agricultural Research Institute, New Delhi. Retrieved from
http:www.iasri.res.in/ebook/EB_SMAR/e-book_pdf%20files/manual%20IV/6-
Geostatistics.pdf (Last accessed 5 February 2011).

Sener, E., Sener, S., and Davraz, A., (2009), Assessment Of Aquifer Vulnerability Based On GIS
And DRASTIC Methods: A Case Study Of The Senirkent-Uluborlu Basin (Isparta,
Turkey), Journal of Hydrogeology 17(2009), 2023-2035.

Thirumalaivasan, D., Karmegam, M., and Venugopal, K., (2003), AHP-DRASTIC: Software For
Spesific Aquifer Vulnerability Assessment Using DRASTIC Model And GIS, Journal of
Modelling and Software 18(2003), 645-656.

View publication stats