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

net/publication/257471634

Assessment of groundwater vulnerability based on a modified DRASTIC

model, GIS and an analytic hierarchy process (AHP) method: The case of
Egirdir Lake basin (Isparta, Turkey)

Article in Hydrogeology Journal · May 2012

DOI: 10.1007/s10040-012-0947-y

CITATIONS READS

271 1,856

2 authors:

Erhan Şener Aysen Davraz

Süleyman Demirel University Süleyman Demirel University
85 PUBLICATIONS 3,679 CITATIONS 85 PUBLICATIONS 3,206 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Aysen Davraz on 04 August 2014.

The user has requested enhancement of the downloaded file.

Assessment of groundwater vulnerability based on a modified
DRASTIC model, GIS and an analytic hierarchy process (AHP) method:
the case of Egirdir Lake basin (Isparta, Turkey)

Erhan Sener & Aysen Davraz

Abstract A DRASTIC-model method based on a Keywords Groundwater vulnerability . Geographic

geographic information system (GIS) was used to study information systems . Analytic hierarchy process .
groundwater vulnerability in Egirdir Lake basin Contamination . Turkey
(Isparta, Turkey), an alluvial area that has suffered
agricultural pollution. ‘Lineament’ and ‘land use’ were
added to the DRASTIC parameters, and an analytic Introduction
hierarchy process (AHP) method determined the rating
coefficients of each parameter. The effect of lineament Groundwater is the most important source of water supply
and land-use parameters on the resulting vulnerability in arid and semi-arid regions due to it large volumes and
maps was determined with a single-parameter sensitiv- its low vulnerability to pollution when compared to
ity analysis. Of the DRASTIC parameters, land use surface waters (USEPA 1985). Due to general high
affects the aquifer vulnerability map most and lineament population growth and industrialization, greater amounts
affects it least, after topography. A simple linear of domestic and industrial effluents are being discharged,
regression analysis assessed the statistical relation which has lead to the pollution of groundwater (Rahman
between groundwater nitrate concentration and the 2008). There are several types of pollutants that appear to
aquifer vulnerability areas; the highest R2 value was predominate in groundwater such as heavy metals,
obtained with the modified-DRASTIC-AHP method. nutrients, pesticides and other organic chemicals, and
The DRASTIC vulnerability map shows that only the fertilizers. Leaching of various pollutants through the
shoreline of Egirdir Lake and the alluvium units have vadose zone gives rise to contamination. Leaching
high contamination potential. In this respect, the processes vary from one location to another (Baalousha
modified DRASTIC vulnerability map is quite similar. 2006; Sener et al. 2009).
According to the modified-DRASTIC-AHP method, the Preventing groundwater pollution is necessary for
lakeshore areas of Senirkent-Uluborlu and Hoyran effective groundwater resource management and ground-
plains, and all of the Yalvaç-Gelendost plain, have high water-vulnerability assessment is important for such
contamination potential. Analyses confirm that ground- groundwater protection. Vulnerability assessment methods
water nitrate content is high in these areas. By divide a geographical area into subareas in terms of its
comparison, the modified-DRASTIC-AHP method has susceptibility to groundwater contamination; then, in areas
provided more valid results. prone to contamination, effective groundwater protection
measures should be carried out (Guo et al. 2007). Two
types of vulnerability are recognized in literature: intrinsic
(or natural) and specific (or integrated) vulnerability.
Received: 22 December 2011 / Accepted: 3 December 2012
Published online: 29 December 2012 Intrinsic vulnerability is a term used to define the
vulnerability of groundwater to contaminants generated
* Springer-Verlag Berlin Heidelberg 2012 by human activities taking into consideration the inherent
geological, hydrological, hydrogeological and hydrogeo-
E. Sener ()) chemical characteristics of an area. Specific vulnerability
Remote Sensing Center, is used to define the vulnerability of groundwater to
Suleyman Demirel University, particular contaminants taking into consideration the
32260 Isparta, Turkey
e-mail: [email protected] contaminant properties and their relationship with the
various components of intrinsic vulnerability (Hamerlinck
A. Davraz and Arneson 1998; Doerfliger et al. 1999; Gogu and
Department of Geological Engineering,
Suleyman Demirel University, Dassargues 2000; Varol and Davraz 2010).
32260 Isparta, Turkey In general, overlay and index methods are quite
e-mail: [email protected] effective to determine groundwater vulnerability and these

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

702
methods are particularly suitable for use with geographic Egirdir Lake basin, and it was determined that the
information systems (GIS), since they usually involve the obtained results are realistic and representative of the
overlaying and aggregation of multiple maps (Tilahun and actual situation in the field (Sener et al. 2009).
Merkel 2010). An overlay and index method is a multi- In this paper, three different groundwater vulnerabil-
criteria model that aggregates the hydrogeological factors ity methods based on GIS are compared. Essentially,
that control the migration of pollutants into the aquifer. It the study aims to achieve this in five steps: (1) to create
combines factors controlling the movement of pollutants a vulnerability map of the study area using the known
from the ground surface into the saturated zone resulting DRASTIC method; (2) to create a vulnerability map of
in vulnerability indices at different locations. The main the study area using a modified DRASTIC method and
advantage is that some of the factors such as rainfall and to determine the effect of the added parameters with a
depth to groundwater, can be available over large areas, single-parameter sensitivity analysis; (3) to create a
which makes them suitable for regional scale assessments vulnerability map of the study area using the modified
(Thapinta and Hudak 2003). However, a major drawback DRASTIC and an analytic hierarchy process (AHP)
is the subjectivity in assigning numerical values to the methods; (4) to check the results of the vulnerability
descriptive entities and relative weights for the different maps with nitrate concentrations in groundwater sam-
attributes (Babiker et al. 2005). ples; (5) to compare the results of the vulnerability
There has been rapid development of groundwater maps prepared using the three different methods, i.e.
vulnerability assessment in the past 10 years, as well as DRASTIC, modified DRASTIC, and modified
the introduction of various new techniques and methods DRASTIC with AHP.
applied to the assessment (Meinardi et al. 1995; Secunda
et al. 1998; Lasserrea et al. 1999; Al-Adamat et al. 2003;
Lake et al. 2003; Thapinta and Hudak 2003; Zhou et al.
2010). One of the most widely used standard groundwater Materials and methodology
vulnerability methods is DRASTIC, developed by the
United States Environmental Protection Agency (USEPA) Location
as a method for assessing groundwater pollution potential. The Egirdir Lake catchment area is located within the
This method uses seven parameters in its calculation of a Lake District in the southwest of Turkey and covers an
‘vulnerability index’. Some researchers have tried to area of 3,417 km2 (Fig. 1). The population of the study
correlate the vulnerability index with contaminant param- area is approximate 122,700. The total population in rural
eters and/or have controlled the assessment with sensitiv- areas of the study area was 66,963 in 2008 (census data),
ity analyses (Kalinski et al. 1994; Rupert 1999; Mclay et which is approximately 55 % of the total population
al. 2001; Javadi et al. 2011). Also, in recent years, the (Sener 2010); whereas the total urban population in 2008
DRASTIC method has been modified by using additional was 55,739, which is approximately 45 % of the total
parameters and/or by ignoring the existing parameters population.
according to the characteristics of the study area (Umar et The Egirdir Lake is the second largest freshwater
al. 2009; Lee 2003; Simsek et al. 2006; Wang et al. 2007; lake in Turkey which is used for different purposes
Guo et al. 2007; Martínez-Bastida et al. 2010; Awawdeh such as irrigation, tourism, and fishing, and for
and Jaradat 2010). supplying the drinking water needs of the city of
The Egirdir Lake basin (Isparta, Turkey) has been Isparta. The groundwater flow direction of the alluvi-
selected as a study area. There are two groundwater um aquifer is toward the Egirdir Lake in the basin. In
basins, Uluborlu-Senirkent and Yalvaç-Gelendost, with the west of the lake, the amount of groundwater
a combined area of 525 km2. Unconfined alluvium is discharge has been calculated as 7.8 hm3/year (Seyman
the main aquifer in the study area, and pollution 2005). The amount of groundwater discharge to the
transport occurs mostly within this unit. Limestone east side of Egirdir Lake by means of the karstic
and flysch units are locally situated under the alluvium aquifer was determined as 114 hm3/year using the
unit. Discharge of groundwater and surface water is MODFLOW model (Soyaslan 2004). The most impor-
toward Egirdir Lake, which is located in the middle of tant surface waters flowing through the basin are Pupa,
the basin. In the basin, groundwater quality is affected Hoyran, Yalvaç and Çay streams, which discharge
negatively from point and nonpoint pollution sources 4.68, 3.42, 26.38 and 18.63 m3/s to Egirdir Lake,
such as agricultural activities, wastewaters, insecure respectively (Sener 2010). Agricultural production
landfill etc. However, the most important pollutant takes place in approximately 1,147 km2 in Egirdir
factor is agricultural activity practiced in the overall Lake basin. In these regions, natural and synthetic
catchment. Therefore, it is important to evaluate fertilizers and agricultural pesticides have been used in
groundwater vulnerability in the basin from the point large amounts, and this is the main cause of the
of view of both groundwater protection in the basin and degradation of groundwater quality in the study area
protection of the Egirdir Lake water quality. In previous (Sener et al. 2009). Also, insecure landfill of municipal
studies, the groundwater vulnerability was evaluated wastes on the permeable aquifer units, mining activi-
using the DRASTIC model based on GIS in the ties and uncontrolled discharge of sewage have
Senirkent-Uluborlu plain, which is located within the affected the groundwater quality negatively.

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

 703

Fig. 1 Location map of the study area

Methodology on the aquifer vulnerability. The seven parameters are then

assigned weights ranging from 1 to 5 reflecting their
Step 1 relative importance. These numerical weights and ratings
In the first step, the GIS-based DRASTIC method is used were established using the Delphi technique by Aller et al.
for assessing the vulnerability of the study area. The (1987). The Delphi technique utilizes the practical and
DRASTIC method is based on the concept of the hydro- research experiences of professionals in the area of interest
geological setting that is defined as “a composite to assess levels of risk. Typically, the experts are asked to
description of all the major geologic and hydrologic rate the risk level of certain activities under a set of initial
factors that affect and control the groundwater movement conditions (Rahman 2008).
into, through and out of an area” (Aller et al. 1987). The The DRASTIC model computes the final DRASTIC
acronym DRASTIC stands for the seven parameters used vulnerability index (DVI) by summing the products of the
in the model, which are: Depth to water, net Recharge, ratings and weights for each of the seven factors:
Aquifer media, Soil media, Topography, Impact of vadose
zone and hydraulic Conductivity. The significant media DVI ¼ DrDw þ RrRw þ ArAw þ SrSw þ TrTw þ IrIw þ CrCw
types or classes of each parameter represent the ranges,
which are rated from 1 to 10 based on their relative effect ð1Þ

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

704

Where D, R, A, S, T, I and C are the seven consideration hydrogeological field observations and
parameters and the subscripts r and w are the investigations. Weighting factors of the seven parame-
corresponding ratings and weights, respectively. All ters were determined depending on location properties
parameter maps were prepared in a GIS environment which are shown in Table 1.
and vulnerability classification of the basin was per- An appealing aspect of the DRASTIC model is that it
formed using GIS techniques. Seven data layers were is flexible. Several studies have shown that unimportant
digitized and were converted to raster data sets using DRASTIC parameters may be ignored, and new ones may
ArcGIS. Then, the DVI was computed and the be added (Al-Adamat et al. 2003; Babiker et al. 2005;
vulnerability map of the basin was prepared taking into Fritch et al. 2000; Thirumalaivasan et al. 2003).

Table 1 Rating and weighting values used in the DRASTIC, modified DRASTIC, and modified-DRASTIC-AHP methods
Parameters Sub parameters DRASTIC Modified DRASTIC Modified-DRASTIC AHP
Weight Rating Total Weight Rating Total AHP AHP Total
weight weight weight rating AHP
weight
Groundwater 0–<10 5 10 50 5 10 50 0.251 0.460 0.11546
depth (m) D 10–<20 7 35 7 35 0.292 0.07329
20–<30 5 25 5 25 0.134 0.03363
30–40 3 15 3 15 0.067 0.01682
>40 1 5 1 5 0.047 0.01180
Net recharge 348–476 4 1 4 4 1 4 0.122 0.062 0.00756
(mm/year) R 477–537 3 12 3 12 0.097 0.01183
538–596 5 20 5 20 0.160 0.01952
597–670 7 28 7 28 0.263 0.03209
671–787 9 36 9 36 0.419 0.05112
Aquifer media A Alluvium 3 10 30 3 10 30 0.063 0.358 0.02255
Karstic 9 27 9 27 0.249 0.01569
Pyroclastic units 7 21 7 21 0.172 0.01084
Neogene deposits 6 18 6 18 0.099 0.00624
Ophiolite 3 9 3 9 0.061 0.00384
Flysch 2 6 2 6 0.038 0.00239
Metamorphics 1 3 1 3 0.024 0.00151
Soil media S None 2 9 18 2 9 18 0.032 0.419 0.01341
Gravel 8 16 8 16 0.263 0.00842
Gravel-Sand 6 12 6 12 0.160 0.00512
Sand-clay 2 4 2 4 0.097 0.00310
Clay 1 2 1 2 0.062 0.00198
Topography 0–<2 1 10 10 1 10 10 0.02 0.470 0.00940
(slope°) T 2–<6 9 9 9 9 0.262 0.00524
6–<12 5 5 5 5 0.144 0.00288
12–20 3 3 3 3 0.079 0.00158
>20 1 1 1 1 0.045 0.00090
Impact of vadose Metamorphic 5 1 5 5 1 5 0.202 0.030 0.00606
zone I Ophiolitic 1 5 1 5 0.042 0.00848
Clay 2 10 2 10 0.057 0.01151
Volcanic 5 25 5 25 0.075 0.01515
Clay-sandstone 6 30 6 30 0.112 0.02262
Conglomerate- 8 40 8 40 0.157 0.03171
sandstone
Limestone 9 45 9 45 0.221 0.04464
Sand-gravel 10 50 10 50 0.305 0.06161
Hydraulic <10−7 3 2 6 3 2 6 0.078 0.062 0.00484
conductivity 10−6–10−7 4 12 4 12 0.097 0.00757
(m/s) C 10−5–10−6 6 18 6 18 0.160 0.01248
10−4–10−5 8 24 8 24 0.263 0.02051
10−3–10−4 10 30 10 30 0.419 0.03268
Lineament 0–250 3 9 27 0.066 0.470 0.03102
(distance -m) Lin 250–500 7 21 0.262 0.01729
500–750 5 15 0.144 0.00950
750–1,000 3 9 0.079 0.00521
>1,000 1 3 0.045 0.00297
Land use Lu Garden 5 10 50 0.167 0.386 0.06446
Agricultural fields 8 40 0.247 0.04125
Settlements 7 35 0.167 0.02789
Grassland-heathland 5 25 0.099 0.01653
Forest 3 15 0.061 0.01019
Bare rock 1 5 0.041 0.00685

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

 705
Step 2 modified-DRASTIC and AHP methods with GIS tech-
In the second step, ‘lineament’ and ‘land use’ parameters niques mentioned previously.
were added to the known DRASTIC method parameters AHP can serve as a powerful tool for considering
and the method was modified (Fig. 2). The second complicated problems that involve several interrelated
vulnerability map of the study area was prepared using objectives (Chuang 2001). The weights of specific criteria
the modified DRASTIC method and GIS techniques are established by ranking their importance and suitability
mentioned previously. The effect of these parameters (Sener et al. 2010). Once the hierarchy has been
was determined with a single-parameter sensitivity anal- established, a pairwise-comparison matrix of each element
ysis which was designed to compare the ‘theoretical’ within each level is constructed.
weights with ‘effective’ weights. The effective weight is a Structurally, the hierarchy is broken down into a series
function of value of the single parameter with regard to of pair comparison matrices, and the participants are asked
the other parameters as well as the weight assigned to it by to evaluate the off-diagonal relationship in one half of
the modified-DRASTIC model (Babiker et al. 2005). The each matrix. The 9-point scale used in typical analytic
effective weight of parameters is calculated using the hierarchy studies ranges from 1 (indifference or equal
following equation: importance) to 9 (extreme preference or absolute impor-
tance). This pairwise comparison enables the decision-
W ¼ ðPrPw=DVI Þ  100 ð2Þ maker to evaluate the contribution of each factor to the
objective independently, thereby simplifying the decision-
Where making process (Rezaei-Moghaddam and Karami 2008).
The pairwise comparisons of various criteria were
W Effective weight of the parameter organized into a square matrix. The diagonal elements of
Pr Parameter rating the matrix were 1. The principal eigenvalue and the
Pw Parameter weight corresponding normalized right eigenvector of the compar-
DVI DRASTIC vulnerability index ison matrix gave the relative importance of the criteria being
compared. The elements of the normalized eigenvector were
weighted with respect to the criteria or sub-criteria and rated
Step 3
with respect to the alternatives (Bhushan and Rai 2004). The
In the third step, an analytic hierarchy process (AHP) consistency of the matrix of order n was then evaluated. If
method was used in determining the rating coefficients this consistency index failed to reach a threshold level, then
of each parameter in the modified-DRASTIC method. the answers to comparisons were re-examined. The consis-
This process has been used to compute the ratings and tency index, CI, was calculated as:
weight of all parameters used in the modified-
DRASTIC method in order to change the initial weight l max n
CI ¼ ð3Þ
factors participating in assessing the vulnerability n1
equation. The third DVI was calculated using new
where CI is the consistency index, 1max is the largest or
rating and weight values for each of the nine parameters
principal eigenvalue of the matrix, and n is the order of the
and a new vulnerability map was prepared using the
matrix. This CI can be compared to that of a random
matrix, RI, such that the ratio, CI/ RI, is the consistency
ratio, CR. As a general rule, CR≤0.1 should be main-
tained for the matrix to be consistent. For this study, CR0
0.0324; thus, indicating that a consistent matrix was
formed. In the final stage, the rating of each alternative
was multiplied by the weights of the sub-criteria and
aggregated to determine the local ratings with respect to
each criterion. The local ratings were then multiplied by
the weights of the criteria and aggregated to determine
global ratings (Bhushan and Rai 2004).
Finally, the prepared vulnerability maps were validated
using hydrochemical data and actual nitrate occurrence in
the aquifer. The water samples were taken from 20 wells
and nitrate contents of the waters were determined. Also,
simple linear regression analysis was made for determi-
nation of the statistical relation between nitrate concen-
tration of groundwater and aquifer vulnerability maps
prepared using the DRASTIC, the modified DRASTIC
and the modified-DRASTIC-AHP methods. However, the
Fig. 2 Schematic representation of the parameters of the modified minimum and maximum DVI values of the three different
DRASTIC method vulnerability maps are not similar; hence, database

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

706
normalization was performed for the maps. The goal of contains only one independent (explanatory) variable, Xi,
database normalization is to decompose relations with for i01,…, n subjects, and is linear with respect to both
anomalies in order to produce smaller, well-structured the regression parameters and the dependent variable. The
relations. The DVI data of the maps were converted to corresponding dependent (outcome) variable is labeled.
between 0 and 100 using the following equation: The model is expressed as

X  Xmin Yi ¼ a þ bXi þ ei ð5Þ

Xnorm ¼  100 ð4Þ
Xmax  Xmin
Where, the regression parameter a is the intercept (on
Where, the y axis), and the regression parameter b is the slope of
the regression line. The random error term ei is assumed
Xnorm Normalized data to be uncorrelated, with a mean of 0 and constant variance
Xmin Minimum index value (Zou et al. 2003). The flowchart of the methodology and
Xmax Maximum index value used methods is shown in Fig. 3.

The purpose of correlation analysis is to measure and

interpret the strength of a linear or nonlinear relationship Results and discussion
between two continuous variables. Pearson correlation
coefficients take on values between −1 and +1, ranging Assessment of aquifer vulnerability with the DRASTIC
from being negatively correlated (−1) to uncorrelated (0) method
to positively correlated (+1). The sign of the correlation The seven hydrogeological parameter (depth to water, net
coefficient (i.e., positive or negative) defines the direction recharge, aquifer media, soil media, topography, impact of
of the relationship. vadose zone and hydraulic Conductivity) should be
Simple regression analysis evaluates the relative impact considered when determining aquifer vulnerability with
of a predictor variable on a particular outcome. This is the DRASTIC method. These input data were obtained
different from a correlation analysis, where the purpose is from literature, field studies, the State Hydraulic Works,
to examine the strength and direction of the relationship the State Meteorology Works and the Management of
between two random variables. A simple regression model Agriculture and Village Works. To carry out the aquifer

Fig. 3 The flowchart of the methodology

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

 707
vulnerability analysis using DRASTIC, seven thematic Village Works and was digitized for use in ArcGIS.
maps were prepared using these input data based on The study area was classified into five groups—clay,
ArcGIS and explained in the following: sandy clay, gravels, gravel-sand and rocky—accord-
ing to soil type. The clays can decrease relative soil
Depth to water table (D). This is an important factor permeability and restrict contaminant migration;
because it determines the depth of materials through hence, the class of clay has the lowest rating value.
which contaminants must pass before reaching the The rating and weight values of this parameter are
water table. It also affects the time available for shown in Table 1.
contamination to undergo chemical and biological Topography (T). Topography is described in the form of
reactions such as dispersion, oxidation, natural atten- slope. Slope degree has quite a high significance, as it
uation, sorption etc. (Ckakraborty et al. 2007). Hence, determines the extent of runoff of the pollutant and the
the greater the depth to the water table the lesser the amount of settling a pollutant might experience before
chance of pollutants arriving at the water table and the infiltrating the soil. A digital elevation model (DEM) of
greater for chance for pollutants to be attenuated. Data the study area was prepared using digitized 1:25,000
from 20 piezometers were used to prepare the scale topographical maps (contour interval of 10 m).
groundwater depth map. According to data collected, Five slope classes were designated to apply to the
the groundwater depth varies from 3 to 36 m in the DRASTIC method. Each class was assigned with a
Senirkent-Uluborlu basin, 1–21.6 m in the Yalvaç rating from 10 (>20°) to 1 (0–2°, Table 1).
basin, 0.15–51.2 m in the Gelendost basin, and 2.6– Impact of vadose zone (I). The vadose zone (unsatu-
38.7 m in the Hoyran basin. The groundwater depth rated zone) has an important role in the percolation of
map of the study area classified areas into five groups rainfall and in surface-water flow. Data on the vadose
(0–<10, 10–<20, 20–<30, 30–40, >40 m). The zone were extracted from the logs and boreholes
groundwater depth weight is determined as 5. The provided by the State Hydraulic Works (SHW) and
rating values of the groundwater depth classes vary the geology map of the study area. Metamorphic,
from 1 to 10. The values for weight and rating are ophiolitic, clay, volcanic, clay-sandstone, conglomer-
presented in Table 1. ate-sandstone, limestone and sand-gravel are the main
Net recharge (R). Net recharge represents the total classes of the vadose zone observed in the basin.
quantity of water applied to the ground surface through These units are evaluated by the DRASTIC method;
precipitation and infiltration to the aquifer. The higher the highest weight value was given in the sand-gravel
the net recharge is, the less the vulnerability to the areas. The rating and weight values of this parameter
aquifer. Annual rainfall data were obtained from the are shown in Table 1.
State Meteorology Works. The rainfall map of the basin Hydraulic conductivity (C). Aquifer hydraulic conduc-
was prepared using these data; rainfall varies from 348 tivity is the ability of the aquifer formation to transmit
to 788 mm in the study area. The ratings for net recharge water. Contamination is controlled by the rate at which
are presented in Table 1. groundwater flows. Based on pumping tests of wells in
Aquifer media (A). Groundwater flow, contaminant the study area, hydraulic conductivity data were
fate and transport modeling are important components obtained. Hydraulic conductivity determined in the west
of most aquifer remediation studies (Rahman 2008). of the study area varies between 8.72×10−6 (m/s) and
The aquifer map of the basin was prepared taking into 2.24×10−4 ( m/s) in alluvium. Hydraulic conductivity
account hydrogeological properties of the lithologic was calculated as 1.18×10−5to 5.6×10−7 m/s in Hoyran
units. Seven hydrogeologic units were determined in plain. Limestone has high permeability (10−3 m2),
the study area. Alluvium and slope deposits were whereas permeability of clayey units is 10−9–10−10 m2.
classified as alluvium aquifer, which is the most Hydraulic conductivity values of the basin have also
vulnerable unit in the basin. Dolomite and limestone been evaluated in five different classes. The rating and
units are karstic rocks and have high permeability. weight values of this parameter are shown in Table 1.
Pyroclastic units and Neogene deposits were identi-
fied as semi-permeable units assigned with a rating of The DRASTIC vulnerability index was computed
7 and 6, respectively. Ophiolite complex, flysch and according to Eq. (1) and is between 60 and 186. Also,
metamorphic rocks were identified as impermeable the DRASTIC vulnerability map of the basin was
units due to having low permeability. According to prepared using overlay analyses of the seven hydro-
DRASTIC standards, the rating of aquifer media geological parameter maps mentioned previously.
varies between 1 for metamorphics and 10 for According to the results of the groundwater vulnerabil-
alluvium aquifer (Table 1). ity assessment, the shoreline of Egirdir Lake and the
Soil media (S). The type of soil cover has a significant alluvium units generally have high contamination
impact on the recharge of the aquifers and it controls potential. Additionally, carbonate units in the study
the movement of contaminants in the vadose zone, area were identified as having medium contamination
making this parameter quite important in determining potential, while the other areas located in the north,
aquifer vulnerability. A hardcopy of the soil map was south and west of the basin have low contamination
obtained from the Management of Agriculture and potential (Fig. 4).

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

708

Fig. 4 Aquifer vulnerability map using the DRASTIC method

Assessment of aquifer vulnerability with the modified Buffer zones of 250 m were drawn around all lineaments
DRASTIC method in the basin and converted into a grid map with 30×30 m
To modify the known DRASTIC method, Lineament and cell sizes, whereby each zone was weighted. The highest
Land use parameters were added to the assessment and, grade (9) was assigned to a buffer zone 0–250 m because
thus, aquifer vulnerability in the study area was re- of high contamination potential. The rating and weight
evaluated. DRASTIC parameters are described in detail values of this parameter are shown in Table 1.
in the preceding and the rating and weight values of these Land use (Lu). This is an important parameter for
were used just the same. Weighting factors of the assessment of groundwater vulnerability. The pollutants
lineament and land use parameters were determined and originating from residential areas and agricultural activ-
used in the vulnerability mapping (Table 1). These added ities could adversely affect the quality of groundwater;
parameters are explained in the following: therefore, a land-use map of the basin was prepared to
evaluate the groundwater contamination potential. The
Lineaments (Lin). Lineaments indicate movement and basin was classified as garden fruits, agricultural fields,
storage of groundwater (Subba Rao et al. 2001) and settlements, grassland, heathland forest and bare rock
therefore are important guides for groundwater explora- according to land-use properties. The DRASTIC weight
tion (Sener et al. 2005). ‘Lineament’ refers to numerous of the land-use parameter was given as 5, and the highest
linear features of the land surface, related to geological grade (10) was assigned to garden-fruits areas because of
evolution, and it is closely related to groundwater flow the high potential impact. The scheme for the grading
and contaminant migration; therefore, higher lineament criterion is shown in Table 1.
density values may mean more potential groundwater
contamination (Lee 2003). Lineament analysis of the The modified-DRASTIC vulnerability map of the basin
basin was carried out using ASTER satellite imagery and was prepared using overlay analyses of the depth to water,
the digital elevation model of the study area (Fig. 5). net recharge, aquifer media, soil media, topography, impact

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

 709

Fig. 5 Lineament map of the study area

of vadose zone, hydraulic conductivity, lineament and land second rating value was determined. The rating
use parameters as already mentioned (Fig. 6). The calculated coefficients used in the preparation of maps by the
DVI value for the modified-DRASTIC map is between 71 modified-DRASTIC-AHP method were calculated by
and 256; high index values correspond to areas with high multiplying the rating values of each parameter with
potential of vulnerability. Accordingly, when the modified- the obtained new rating value (Table 1). The modified-
DRASTIC vulnerability map was compared with the DRASTIC-AHP vulnerability map of the basin was
DRASTIC vulnerability map, similar results were obtained prepared using overlay analyses of the mentioned nine
with both maps and almost the same areas in the basin had parameters, as for the other methods (Fig. 7). When
the same high, medium, and low-contamination potential. the modified-DRASTIC-AHP vulnerability map was
compared with the other vulnerability maps obtained
from the DRASTIC and modified DRASTIC methods,
Assessment of aquifer vulnerability the classes of vulnerability were defined more clearly
with the modified-DRASTIC-AHP method with the use of weight values determined by the AHP
In order to put forward a more realistic method to classify method. The obtained vulnerability index value is
the groundwater vulnerability, the ranges of the parame- between 0.3195 and 0.0687 with this method.
ters of the modified DRASTIC model were modified using According to the modified-DRASTIC-AHP vulnerabil-
an analytic hierarchy process (AHP). The pairwise- ity map, shore areas of the Senirkent-Uluborlu and
comparison matrix was prepared for the nine parame- Hoyran plains, and all of the Yalvaç-Gelendost plain,
ters used in the modified DRASTIC method and new have high contamination potential. Low contamination
rating coefficients were calculated for each parameter. potential was determined in the areas in which clay
Then, each parameter was re-evaluated in itself and the and metamorphic units overlay. Additionally, as in the

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

710

Fig. 6 Aquifer vulnerability map using the modified DRASTIC method

other methods, carbonate units were identified as The theoretical and effective weights calculated with the
having medium contamination potential (Fig. 7). SPSA of the parameters can be shown in Table 2. In the
modified DRASTIC method, average effective weights of
the impact of vadose zone, aquifer media and land-use
parameters were calculated as 23.0, 13.2 and 16.08 %,
Single-parameter sensitivity analysis (SPSA) respectively (Table 2). As a result of the SPSA, the land-use
The main objective of single-parameter sensitivity analy- parameter has higher effective weight than the lineament
sis (SPSA) is to determine the weight of the ‘effective parameter in the aquifer vulnerability map.
effect’ of each parameter in the aquifer vulnerability map.
In this way, the ‘most effective impact’ parameters are
determined by comparing the theoretical effect weight with
the effective effect weight of each parameter. In this section Validation of the vulnerability maps
of the study, the SPSA was applied to the aquifer The DRASTIC method considered that pollutants have the
vulnerability maps obtained using the modified DRASTIC same fluidity as water. Nitrate fertilizers are used in the study
method. Thus, effective effect weights of the lineament and area, and nitrate compounds have high solubility potential in
land-use parameters were compared with the other param- the water. Hence, the validation of the vulnerability maps
eters. Firstly, the minimum, maximum, mean and standard was controlled with nitrate concentrations in groundwater
deviation values of the hydrogeological parameter maps samples taken from 20 locations in the basin. The nitrate
were determined. Accordingly, while the impact of the concentration in groundwater near the Egirdir Lake shore
vadose zone parameter has the highest average, topography was more than 30 mg/L during May 2010. A nitrate
(T) has the lowest average value and the average value of the distribution map of the basin was prepared using the analysis
lineament and land use parameters were determined as results and compared with the vulnerability maps. In general,
approximately 11 and 22.82, respectively (Table 2). nitrate concentration gradually increases towards the lake

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

 711

Fig. 7 Aquifer vulnerability map using the modified-DRASTIC-AHP method

depending on groundwater flow direction and agricultural groundwater nitrate content was used as the dependent
applications, and these areas have high contamination variable in the SLRA. According to SLRA results, the R2
potentials (Figs. 4, 6, 7). values of the vulnerability maps obtained from the
To determine the statistical relation between nitrate DRASTIC, modified DRASTIC and modified-
concentration of groundwater and aquifer vulnerability DRASTIC-AHP methods were 0.474, 0.430 and 0.490,
maps, SLRA was undertaken after the database normali- respectively (Fig. 8a–c). Finally, the vulnerability map
zation was performed on the maps. The cell values prepared using the modified-DRASTIC-AHP method was
(corresponding sampling location in the vulnerability found to have higher accuracy than the other methods, on
maps) were used as independent variables, and the account of the highest R2 value.

Table 2 Statistical analysis of the effective weight. SD standard deviation

Parameter Theoretical weighting Theoretical weighting (%) Effective weighting (%)
Min. Max. Average SD
D 5 16.129 2.2 40.0 6.090 6.971
R 4 12.903 1.7 34.9 10.950 5.603
A 3 9.677 1.5 36.9 13.208 4.885
S 2 6.452 0.8 21.9 7.743 4.220
T 1 3.226 0.4 13.3 3.172 2.087
I 5 16.129 2.5 45.0 23.006 9.116
C 3 9.677 2.4 42.2 11.705 7.546
Lin 3 9.677 1.2 28.4 7.621 5.913
Lu 5 16.129 0.0 39.0 16.081 8.202

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

712

Fig. 8 The relationship between nitrate concentration and aquifer vulnerability maps: a DRASTIC method, b modified DRASTIC
method, c modified-DRASTIC-AHP method

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

 713

Conclusions Also, the lineament parameter shows lower effective

weights when compared with the land-use parameter. In
In this study, the DRASTIC method was used for addition, the aquifer vulnerability maps that were prepared
evaluation of groundwater vulnerability in the Egirdir using the different methods were compared with nitrate
Lake basin. Groundwater vulnerability assessments of the concentrations of the groundwater. Generally, nitrate con-
entire basin are found to be realistic and suitable for the centrations in groundwater from the areas that are classified
Senirkent-Uluborlu plain. However, despite its success in as highly vulnerable are measured at high concentrations.
some case studies, the DRASTIC method has some Furthermore, simple linear regression analysis (SLRA) was
disadvantages. The influence of regional characteristics undertaken to determine the statistical relation between the
(geology, hydrology, hydrogeology, land use etc.) is not nitrate concentration of groundwater and the aquifer vulner-
accounted for in the method and so the same weights and ability maps. According to this evaluation, the modified-
rating values are used everywhere. If the detail hydro- DRASTIC-AHP method has higher accuracy than the other
geological properties such as aquifer type, aquifer thick- methods, with 49 % accuracy. For this reason, it can be
ness, groundwater level and groundwater flow direction concluded that the modified-DRASTIC-AHP method gave
etc., are well known, more reliable results can be obtained more valid and accurate results than other methods for the
with this method. In order to better address local issues for alluvial aquifer vulnerability assessments.
refined representation of local hydrogeologic settings, The groundwater vulnerability maps of the Egirdir Lake
researchers envisaged several modifications of the original catchment area are ideal for use in future land-use planning
DRASTIC model. The modifications were in the form of studies. Detailed and frequent groundwater quality monitor-
additional parameters, removal of certain parameters and ing in highly vulnerable areas should be performed to
usage of different ratings and weights for the parameters. monitor changing levels of pollutants. In addition, local
In this study, the DRASTIC method was modified with administrators should conduct education programs and
added lineament and land-use parameters. The effect of awareness-raising activities for the farmers.
contamination on groundwater quality is greatest in areas
with dense lineaments. Hence, the lineament parameter Acknowledgements This study was supported by The Scientific and
was added, because dense areas of lineaments have been Technological Research Council of Turkey (TUBITAK) along with
project No. 108Y258 and Scientific Research Projects of the Suleyman
observed in the study area. Land use is one of the Demirel University (project No. 81-DES-09). The authors would like
important parameters reflecting anthropogenic impact. The to thank Scientific Research Projects of Suleyman Demirel University
calculated DVI value for the modified-DRASTIC map is which supported this work with the MSc Thesis project (project No.
between 71 and 256; and, in general, the shores of Egirdir 2119). In addition, the authors would like to thank Assistant Prof. Dr.
Lake, and alluvium units, have high-contamination poten- Şehnaz Şener for her contribution, and also Sue Duncan for the careful
technical editing.
tial. Additionally, carbonate units in the study area were
identified as having medium-contamination potential. The
other areas in the north, south and west of the basin have
low-contamination potential. Similar results were obtained References
from the DRASTIC and the modified-DRASTIC methods.
The accuracy of weights and rating values of the Al-Adamat RAN, Foster IDL, Baban SMJ (2003) Groundwater
DRASTIC parameters is important for validity of this vulnerability and risk mapping for the basaltic aquifer of the
method. The DRASTIC method is rigid in the assignment Azraq Basin of Jordan using GIS, remote sensing and
DRASTIC. Appl Geogr 23:303–324
of ratings and weights to the model parameters. In this Aller L, Bennet T, Leher JH, Petty RJ, Hackett G (1987) DRASTIC:
study, detailed hydrogeological field studies were per- a standardized system for evaluating groundwater pollution
formed to reduce the margin of error. The rating values of potential using hydro-geological settings. EPA 600/2-87-035,
the parameters were determined, dependent on field and USEPA, Washington, DC, 622 pp
Awawdeh MM, Jaradat RA (2010) Evaluation of aquifers vulner-
regional properties of the study area. Furthermore, the ability to contamination in the Yarmouk River basin, Jordan,
analytic hierarchy process (AHP) method was used in based on DRASTIC method. Arab J Geosci 3:273–282
determining the rating coefficients of each parameter in Baalousha H (2006) Vulnerability assessment for the Gaza Strip,
the modified-DRASTIC method. The vulnerability classes Palestine using DRASTIC. Environ Geol 50:405–414
of the study area were defined more clearly using the Babiker IS, Mohammed MAA, Hiyama T, Kato K (2005) A GIS-
based DRATIC model for assessing aquifer vulnerability in
modified-DRASTIC-AHP method. Shoreline areas of Kakamigahara Heights, Gifu Prefecture, central Japan. Sci Total
Senirkent-Uluborlu and Hoyran plains, and all of the Environ 345:127–140
Yalvaç-Gelendost plain, have high contamination potential Bhushan N, Rai K (2004) Strategic decision making: applying the
according to the vulnerability map that was prepared using analytic hierarchy process. Springer, New York
Chuang PT (2001) Combining the analytic hierarchy process and
the modified-DRASTIC-AHP method. quality function deployment for a location decision from a
The effects of the DRASTIC, lineament and land-use requirement perspective. Int J Adv Manuf Technol 18:842–849
parameters on the vulnerability maps were investigated Ckakraborty S, Paul PK, Sikdar PK (2007) Assessing aquifer
using a single-parameter sensitivity analysis (SPSA). vulnerability to arsenic pollution using DRASTIC and GIS of
North Bengal Plain: a case study of English Bazar Block, Malda
According to SPSA results, Impact of vadose zone, District, West Bengal, India. J Spat Hydrol 7(1):101–121
aquifer media and land-use parameters have the highest Doerfliger N, Jeannin PY, Zwahlen F (1999) Water vulnerability
effective weights compared with the other parameters. assessment in karst environments: a new method of defining

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

714
protection areas using a multi-attribute approach and GIS tools Sener E, Davraz A, Ozcelik M (2005) An integration of GIS and
(EPIK method). Environ Geol 39(2):165–176 remote sensing in groundwater investigations: a case study in
Fritch TG, McKnight CL, Yelderman JC, Arnold JG (2000) Aquifer Burdur, Turkey. Hydrogeol J 13(5–6):826–834
vulnerability assessment of the Paluxy Aquifer, central Texas, Sener E, Sener S, Davraz A (2009) Assessment of aquifer vulnerability
USA, using GIS and a modified DRASTIC approach. Environ based on GIS and DRASTIC methods: a case study of Senirkent-
Manag 25:337–345 Uluborlu basin (Isparta, Turkey). Hydrogeol J 17:2023–2035
Gogu RC, Dassargues A (2000) Current trends and future Sener S (2010) Hydrogeochemical investigation of Egirdir Lake
challenges in groundwater vulnerability assessment using water and bottom sediments (in Turkish). PhD Thesis, Suleyman
overlay and index methods. Environ Geol 39(6):549–558 Demirel University, Turkey
Guo O, Wnag Y, Gao X, Ma T (2007) A new model (DRARCH) for Sener S, Sener E, Nas B, Karagüzel R (2010) Combining AHP with
assessing groundwater vulnerability to arsenic contamination at GIS for landfill site selection: a case study in the Lake Beyşehir
basin scale: a case study in Taiyuan basin, northern China. catchment area (Konya, Turkey). Waste Manage. doi:10.1016/
Environ Geol 52:923–932 j.wasman.2010.05.024
Hamerlinck JD, Arneson CS (1998) Wyoming groundwater Seyman F (2005) Hydrogeological investigations of Senirkent-
vulnerability assessment handbook, vol 1: background, model Uluborlu (Isparta) basin (in Turkish). Suleyman Demirel
development and aquifer sensitivity analysis. SDVC 98-01-1, University, Master Thesis, Turkey
Spatial Data and Visualization Center, Laramie, WY Soyaslan II (2004) Hydrogeological investigations of eastern
Javadi S, Kavehkar N, Mousavizadeh MH, Mohammadi K (2011) Egirdir Lake and groundwater modeling (in Turkish). PhD
Modification of DRASTIC model to map groundwater vulner- Thesis, Suleyman Demirel University, Turkey
ability to pollution using nitrate measurements in agricultural Simsek C, Kincal C, Gunduz O (2006) A solid waste disposal site
areas. J Agric Sci Tech 13:239–249 selection procedure based on groundwater vulnerability map-
Kalinski RJ, Kelly WE, Bogardi I, Ehrman RL, Yamamoto PO ping. Environ Geol 49:620–633
(1994) Correlation between DRASTIC vulnerabilities and Subba Rao N, Chakradhar GKJ, Srinivas V (2001) Identification of
incidents of VOC contamination of municipal wells in groundwater potential zones using remote sensing techniques in
Nebraska. Ground Water 32(1):31–34 and around Guntur Town, Andhra Pradesh, India. J Indian Soc
Lake IR, Lovett AA, Hiscock KM, Betson M, Foley A, Sunnenberg Remote Sensing 29(1–2):69–78
G, Evers S, Fletcher S (2003) Evaluating factors influencing Thapinta A, Hudak PF (2003) Use of geographic information
groundwater vulnerability to nitrate pollution: developing the systems for assessing groundwater pollution potential by
potential of GIS. J Environ Manag 68:315–328 pesticides in central Thailand. Environ Int 29(1):87–93
Lasserrea F, Razacka M, Bantonb O (1999) A GIS-linked model for Thirumalaivasan D, Karmegam M, Venugopal K (2003) AHP-DRAS-
the assessment of nitrate contamination in groundwater. J TIC: software for specific aquifer vulnerability assessment using
Hydrol 224:81–90 DRASTIC model and GIS. Environ Model Softw 18:645–656
Lee S (2003) Evaluation of waste disposal site using the DRASTIC Tilahun K, Merkel BJ (2010) Assessment of groundwater vulner-
system in southern Korea. Environ Geol 44:654–664 ability to pollution in Dire Dawa, Ethiopia using DRASTIC.
Martínez-Bastida JJ, Arauzo M, Valladolid M (2010) Intrinsic and Environ Earth Sci 59:1485–1496
specific vulnerability of groundwater in central Spain: the risk Umar R, Ahmed I, Alam F (2009) Mapping groundwater vulnerable
of nitrate pollution. Hydrogeol J 18:681–698 zones using modified DRASTIC approach of an alluvial aquifer
McLay CDA, Dragden R, Sparling G, Selvarajah N (2001) in parts of Central Ganga Plain, Western Uttar Pradesh. J Geol
Predicting groundwater nitrate concentrations in a region of Soc India 73(2):193–201
mixed agricultural land use: a comparison of three approaches. United States Environmental Protection Agency, USEPA (1985)
Environ Pollut 115:191–204 DRASTIC: a standard system for evaluating groundwater
Meinardi CR, Beusen AHW, Bollen MJS, Klepper O, Willems WJ potential using hydrogeological settings. WA/EPA Series
(1995) Vulnerability to diffuse pollution and average nitrate 1985, US EPA, Washington, DC, 163, pp
contamination of European soils and groundwater. Water Sci Varol S, Davraz A (2010) Hydrogeological investigation of
Technol 31:159–165 Sarkikaraagac Basin (Isparta, Turkey) and groundwater vulner-
Rahman A (2008) A GIS based DRASTIC model for assessing ability. Water Int 35(2):177–194
groundwater vulnerability in shallow aquifer in Aligarh, India. Wang Y, Merkel BJ, Li Y, Ye H, Fu S, Ihm D (2007)
Appl Geogr 28:32–53 Vulnerability of groundwater in quaternary aquifers to
Rezaei-Moghaddam K, Karami E (2008) A multiple criteria organic contaminants: a case study in Wuhan City, China.
evaluation of sustainable agricultural development models using Environ Geol 53:479–484
AHP. Environ Dev Sustain 10:407–426 Zhou J, Li G, Liu F, Wang Y, Guo X (2010) DRAV model and
Rupert MG (1999) Improvements to the DRASTIC groundwater its application in assessing groundwater vulnerability in
vulnerability mapping method. US Geol Surv Fact Sheet FS-066-99 arid area: a case study of pore phreatic water in Tarim
Secunda S, Collin ML, Melloul AJ (1998) Groundwater vulnera- Basin, Xinjiang, Northwest China. Environ Earth Sci
bility assessment using a composite model combining DRAS- 60:1055–1063
TIC with extensive agricultural land use in Israel’s Sharon Zou KH, Tuncali K, Silverman SG (2003) Correlation and simple
region. J Environ Manag 54:39–57 linear regression. Radiology 227:617–628

Hydrogeology Journal (2013) 21: 701–714 DOI 10.1007/s10040-012-0947-y

View publication stats