Arab J Geosci

DOI 10.1007/s12517-012-0765-5

ORIGINAL PAPER

Geodetic deformation monitoring of Ataturk Dam in Turkey

Yunus Kalkan

Received: 11 July 2012 / Accepted: 1 November 2012

# Saudi Society for Geosciences 2012

Abstract Dams are among the most important engineering may still be significant during impounding of the reservoir and
structures used for water supply, flood control, agricultural generally diminish with time but time-dependent creep may
uses, and hydroelectric power. Monitoring of dams is crucial continue at a slow rate for several years (Clements 1984;
since deformation might have occurred as a result of ero- Gikas and Sakellario 2008; Kim and Kim 2008). Geodetic
sion, water load, hydraulic gradients, and water saturation. and non-geodetic methods have been used to find out the
This research provides information about the deformation potential deformations that occurred on and around dams
monitoring techniques. A case study was conducted on (Hudnut and Behr 1998; Chrzanowski and Massiera 2004;
Ataturk Dam, the largest dam of Turkey constructed on Fırat Aguilera et al. 2007; Taşçı 2008; Kalkan et al. 2009).
River, to determine the magnitude and the direction of radial Aguilera et al. (2008) proposed the application of terrestrial
deformations that took place between 2006 and 2010. At the laser scanner (TLS) to the structural monitoring of a large dam
end of analyses, decreasing radial movement velocity was in Spain. They considered aspects regarding to the accuracy
determined generally. Monthly average radial velocity of the control in georeferencing, together with rigorous approaches
last period which consists 6 months is 2.2 mm. As a result of to model complex structures. They employed radial basis
the study, no significant correlation was determined between function for the surface parameterization and re-weighted
radial movements on embankment and reservoir water level. extended orthogonal Procrustes analysis for georeferencing.
Their results showed that a TLS sensor alone is not enough to
Keywords Dam . Deformation monitoring . Geodetic provide the structural control of large dams, since it is not
techniques . Ataturk Dam possible to scan the same point in different measurement
periods. However, massive data points could be collected via
TLS technology which makes them unsurpassable.
Introduction Taşçı (2010) established a network of six reference and
11 object points to monitor and analyze the deformations at
There are different factors which could cause failure of the the crest of the Altinkaya Dam. In the study, static GPS
embankment dams like internal erosion, stability problems measurement method and different analysis methods like
due to high pore pressures and hydraulic gradients. Moni- iterative weighted transformation, least absolute sum, and
toring of dam behavior during and after construction is congruency test were used. The result showed that horizon-
important to determine deformations occurring on dams tal movements caused by water load effect could occur in
and to secure and preserve the safety of dams (Chrzanowski the middle of the dam’s crest in arch dams.
et al. 1991; Chrzanowski and Massiera 2004; Aguilera et al. This research determines magnitude and direction of
2008). Despite the compaction efforts, internal displacements deformations on Ataturk Dam, which is the 11th largest
dam of the world, considering its reservoir capacity (Bureau
of Reclamation 2012; Einfopedia 2012), between 2006 and
Y. Kalkan (*) 2010. Deformation measurements were conducted periodi-
Department of Geomatics Engineering, Istanbul Technical
cally (twice a year till 2008, subsequently yearly). GPS and
University, 34469 Maslak,
Istanbul, Turkey conventional geodetic measurement techniques have been
e-mail: kalkany@itu.edu.tr utilized within the study area. Radial movements of the
 Arab J Geosci

Fig. 1 Location of the study area and dam embankment

points located on the embankment were determined and the might cause geometric and physical changes along the dams.
relationship between these movements and reservoir water It is important to determine whether these changes are signif-
level was investigated. icant or they are close to some critical threshold values. Thus,
precautionary measures could be taken to secure and preserve
the safety of dams and to provide the sustainability and effi-
Deformation monitoring ciency of the dams. Both geodetic and non-geodetic methods
have been used to monitor deformations that occurred on dams
Deformations might have occurred on dams and/or their sur- or other engineering structures (USACE 2002; Bilgi et al.
rounding areas because of different factors. Structure of the 2006; Kalkan et al. 2009, 2010).
dam, weight of embankment, water load and pressure, water
pressure changes on the embankment, temperature changes, Geodetic methods
and crustal movements are among the factors which might
cause deformations (Vladimír and Miloš 2004). It is important Geodetic methods are used for the determination of defor-
to monitor dams periodically since the mentioned factors mations. These methods consist from conventional and

Fig. 2 Embankment and

reference network
Arab J Geosci

Fig. 3 Deformation points located on the embankment

space-based techniques such as GPS and interferometry. stable surfaces where deformation is not expected have been
Different equipments and devices have been used on the established and replacements on the dam embankment and
geodetic methods such as theodolites and electronic distance surrounding have been found out with respect to the reference
meter or total stations for conventional measurements, GNSS points. The potential places which might have deformation are
receivers for space-based measurements, and SAR images for represented with object points (deformation points).
interferometric applications (USACE 2002; Aguilera et al. The selection of surveying type, surveying period, and
2007, 2008; Taşçı 2008; Kalkan et al. 2010). related standards for the surveying methods are determined
Deformation monitoring networks have been formed, and considering the type of the dam and magnitude of the expected
conventional and space-based techniques have been con- deformation. Monitoring accuracy for the horizontal displace-
ducted periodically to determine horizontal and vertical defor- ment of concrete arch dams for long-term period measure-
mations on the dams. Reference points outside the region on ments needs to be ±5–10 mm whereas vertical displacement

Fig. 4 GPS measurement

sketch and a GPS point on the
a) b)
dam embankment

1 : 500
 Arab J Geosci

Table 1 Calculated difference vectors and statistical test values

Point dYi dXi dPi MPi dHi MdHi DHi Bearing Constant Difference Used Radial Test
no angle bearing bearing displacement value
angle angle
(cm) (cm) (cm) (cm) (cm) (cm) (cm) ti (grad) ts (grad) ti −ts (grad) dRi (cm) Ti
(cm)

4070 −9.5 −10.5 14.1 0.7 −13.0 0.8 3.0 246.780 252.000 −5.220 394.780 14.1 2.6
2082 −11.4 −8.3 14.1 1.5 −14.0 1.2 4.4 259.720 267.000 −7.280 392.720 14.0 5.3
4050 −11.7 −7.4 13.8 0.9 −13.5 1.2 4.2 263.946 267.000 −3.054 396.946 13.8 3.2
4060 −9.7 −10.0 13.9 0.7 −15.0 0.8 2.8 249.124 263.000 −13.876 386.124 13.6 2.5
4080 −8.6 −10.4 13.5 0.7 −11.3 0.7 2.5 243.903 244.000 −0.097 399.903 13.5 2.4
2112 −8.8 −10.3 13.5 1.4 −10.5 1.1 3.8 245.164 248.000 −2.836 397.164 13.5 5.1
2062 −9.7 −8.2 12.8 0.9 −10.6 0.9 3.4 255.369 267.000 −11.631 388.369 12.5 3.2
2102 −8.6 −9.1 12.5 1.5 −11.3 1.1 3.9 248.239 252.000 −3.761 396.239 12.5 5.1
2101 −8.3 −8.9 12.2 1.4 −8.9 1.1 3.9 247.517 252.000 −4.483 395.517 12.1 5.1
2092 −8.4 −9.1 12.4 1.5 −13.0 1.2 4.2 247.388 262.000 −14.612 385.388 12.1 5.2
2056B −10.8 −4.4 11.7 0.5 −13.0 0.7 2.3 275.524 275.000 0.524 0.524 11.7 1.9
4040 −9.3 −6.6 11.4 0.8 −11.2 0.8 3.0 260.928 267.000 −6.072 393.928 11.4 2.7

dP represents the position vector and dR is the radial component perpendicular to dam embankment.

needs to be ±2 mm. On the other hand, monitoring the accu- Ataturk Dam. The dam is not only the largest dam of Turkey
racy of horizontal displacement should be ±20–30 mm and but also one of the largest dams in world, considering its
vertical displacement should be ±10 mm for the rock fill dams embankment volume (Department of the Interior, Bureau of
(Deloach 1989; USACE 2002). Reclamation and International Water Power and Dam
Construction 2007).
Technical data of Ataturk Dam Ataturk Dam was constructed between 1983 and 1992. It
has been used for irrigation, drinking water, and energy
There are approximately 600 dams in Turkey with various production purposes. The height of the dam is 169 m and
capacities (Özkan and Yılmaz 1999). Ataturk Dam is one of the total volume is 84,500 hm3, whereas the reservoir vol-
the five dams constructed at Fırat River as part of the ume is 48,700 hm3 and the reservoir area is 817 km2.
Southeast Anatolia Project. Figure 1 shows the location of Approximately 900,000 ha area in Harran Plain, Sanliurfa

Fig. 5 Horizontal displacement

vectors (length is in millimeter)
for the points on the
embankment (between May
2006 and November 2010)
Arab J Geosci

employed (Electrowatt Engineering Ltd; Dolsar Engineering

Ltd 2004; Bilgi et al. 2006; Kalkan et al. 2009, 2010).

Deformation monitoring of Ataturk Dam using geodetic

methods

The geodetic measurements of the dam deformations began in

July 1990, just before the completion of the construction (Malla
et al. 2007). A deformation network consisting of reference
points and object/deformation points was used in the study.
Position changes on the dam embankment and surroundings
are defined relatively to the reference points. These points are
established on the areas where deformation would not occur.
A reference network containing 32 points located on
stable ground was established to monitor the deformations
of Ataturk Dam. Seven benchmarks covering the most outer
part of the area were designed as a GPS network, whereas
25 benchmarks inside the GPS network were designed for
conventional surveying (Fig. 2). Concrete pillars were con-
Fig. 6 Comparison of point positions obtained from GPS measure-
structed for GPS and other reference points whereas force-
ment on the embankment (between May 2007 and November 2010) centered structures were used for the object points.
Deformation network consists of 360 benchmarks locat-
ed on the embankment and floats (Fig. 3). GPS and conven-
is irrigated by the water provided by the dam using tunnels tional geodetic techniques were applied for deformation
and pipe stations. Moreover, drinking and usage water of the monitoring of the dam.
city of Sanliurfa is obtained from the Ataturk Dam. The dam
also feeds a hydroelectric power plant which provides a GPS measurements
significant amount of energy to Turkey (Turkish General
Directorate of State Hydrologic Works 2005). Twenty-seven points of reference network (formed by 32
Ataturk Dam has been monitored using geodetic and points) have been measured with GPS technique. The meas-
non-geodetic methods to determine the multi-temporal urements were conducted between May 2006 and November
deformations on the dam. Some measurement points have 2010. The occupation time of measurements varied between 3
been selected and a number of equipments have been placed and 8 h. Data sampling of 5 s and elevation mask of 10° were
to monitor the changes on dam foundation, embankment, used throughout the GPS campaign. Coordinates of approxi-
concrete structures, side berms, power station, and galleries. mately 200 object points on the embankment have also been
In addition to geodetic measurements, non-geodetic meas- determined with GPS measurements (Fig. 4) since the third
urements such as pore water, temperature, slope, displace- period (i.e., May 2007). Technical details of GPS measure-
ment, strain, and crack measurements have also been ments are given by Kalkan et al. (2009, 2010).

Table 2 Radial movement and velocity of point 2,082 with respect to mean water level change of the reservoir

Evaluation time Evaluation period Mean water Mean water Radial Monthly average
level change(m) level (m) displacement radial velocity
(mm) (mm/month)

First 6 months May 2006–Nov 2006 −1.48 535.37 −25 4.2

Second 6 months Nov 2006–May 2007 +0.22 535.39 −17 2.7
Third 6 months May 2007–Nov 2007 −4.05 533.22 −20 3.3
Fourth 6 months Nov 2007–May 2008 −0.53 531.42 −26 4.3
Fifth 6 months May 2008–Nov 2008 −2.62 529.17 −11 1.8
Sixth and seventh 6 months Nov 2008–Nov 2009 −0.17 528.45 −16 1.3
Eigth and ninth 6 months Nov 2009–Nov 2010 +6.94 533.45 −26 2.2
 Arab J Geosci

Conventional measurements Distance measurements were carried out reciprocal for

the reference network points and single for the object points.
Precise horizontal direction and vertical angle and distance These measurements were reduced to the projection surface
measurements were carried out with conventional geodetic applying necessary corrections. After the adjustment of each
method. Coordinates of 17 reference points in the proximity period measurement with appropriate stochastic model,
of the embankment were determined with conventional meth- ±1.6 mm point position accuracy and ±2.3 mm vertical
ods (Fig. 2). In addition, 200 objects points located on the dam accuracy for triangulation network and ±3.5 mm point po-
embankment were measured using nine pillars to relate refer- sition accuracy and ±4.3 mm vertical accuracy for object
ence network and object points with the same methods. network were obtained. Measurement results of eight peri-
The angle measurements were carried out as three sets for ods were compared with each other. In order to conduct
the observations of points within the reference network and double period analysis, stochastic test was applied to each
two sets for the object points. The difference vector for the period to determine whether it could evaluate these two
station adjustment is calculated using the following equation: periods in the same set. Details of the test are given below.
―
X The congruence testing of deformation networks are
di ¼ l i  li s ; vi ¼ di  di =n ð1Þ mainly conducted using the statistical techniques. The
where lis represents direction observations, i represents the results of the testing method is used to decide whether a
direction number (i01, 2, 3, …, n), s represents the number point coordinate or invariant differences are statistically
of series (s01, 2, 3…), and “īi” represents the mean of sets. significant or not (Sedlak and Jecny 2004).
Additionally, root mean square (RMS) error value of a single Definition of the null hypothesis H0 is the first step and it
observation (m0) and RMS of the mean of the observations is given in Eq. 3:
(M) are calculated by the following equation
 X 1=2 .    2
m0 ¼ vi vi =ðn  1Þðs  1Þ ; M ¼ m0 ðnÞ1=2 ð2Þ H0 : E s01 2 ¼ E s0f 2 ¼ σ0 2 ð3Þ

Fig. 7 Mean water level of the reservoir and average radial movement velocity over the dam embankment between May 2006 and November 2010 (For
the points that have the biggest radial movement velocities.)
Arab J Geosci

Where σ0 is the selected standard deviation, and s0i and s0f value for positions (TPi) and test value for heights (THi) were
are mean square error values of unit measure for the first and calculated using the following equations:
  
second period, respectively. Test value of f f ¼ s01 2 s0f 2
was calculated with 0.05 error margin. Considering F(f1,ff) a TPi ¼ 2; 5  ð2Þ0:5  ðMdPi Þ; THi ¼ 2; 5  ð2Þ0:5  ðMdHi Þ ð4Þ
table value, it was calculated whether the conditions (fN <FN)
for triangulation network and (fT <FT) for all network were and these values were used for comparison. If dPi >TPi and dHi >
met or not. When the same set was evaluated, displacement THi, then there is a significant difference for the point of i.
vectors of periods and their accuracy values were calculated The results obtained after the analysis of November 2010
by double period analysis. To determine if obtained difference measurements show that there were significant horizontal
values are significant within 95 % confidence interval, test movements among the 72 % of object points, whereas 71 %

Fig. 8 Radial displacements

(dR) on longitudinal profiles
points at various elevations
within the 54-month period.
Embankment displacement
components perpendicular to
the crest (radial displacements),
+dR, displacement towards
downstream perpendicular to
the crest axis; −dR, displace-
ment towards upstream perpen-
dicular to the crest axis
 Arab J Geosci

of object points were faced with significant radial movements. have the most radial movement velocity. Hence, it is
The magnitude of maximum horizontal movement was found difficult to conclude the radial displacement on the crest
as 14.12 cm (with a radial component of 14.08 cm) consider- depending to the reservoir water level exactly.
ing all measurements conducted in the last 4.5 years. Table 1 The obtained results reveals that total radial displacement
shows some points having the maximum horizontal and ver- of point 2,082 was −151 mm with a 2.8 mm monthly average
tical movements located on the embankment. radial velocity with respect to the water level change of the
Figure 5 shows the displacement vectors of all points located reservoir for 54 months (between May 2006 and November
on the embankment. The positions of points, located on the 2010). Figure 8 shows the radial displacement graphics of the
embankment, were obtained using GPS measurements in dif- profiles parallel to the crest which has different height values.
ferent periods. Horizontal displacement vectors of these points In addition, Fig. 9 shows the radial displacement graphics of
between May 2007 (period 3) and November 2010 (period 8) six cross sections perpendicular to the crest.
were calculated by comparing GPS-derived positions (Fig. 6).
The largest movements have occurred on points 2,081,
2,082, 2,083, and 2,084 which are close to 0+230 cross Conclusions
section on the embankment. The location of this section and
points are presented in Fig. 3. Table 2 shows the radial Ataturk Dam has been monitored using geodetic and non-
movement and its velocity of point 2,082 with respect to the geodetic techniques since 1990. The results of geodetic
mean water level change of the reservoir. Additionally, Fig. 7 measurements conducted between May 2006 and November
shows the radial velocity graphs of the points which 2010 were evaluated in this study. Our analyses showed that
Fig. 9 Radial displacements on
cross-sectional points (1-1, 2-2,
3-3, 4-4, 5-5, and 6-6) within
the 54-month period

(1 – 1) Section, km (0 – 480) (2 – 2) Section, km (0 – 230)

(3 – 3) Section, km (0 + 000) (4 – 4) Section, km (0 + 230)

(5 – 5) Section, km (0 + 460) (6 – 6) Section, km (0 + 690)

Arab J Geosci

significant horizontal displacements occurred on 72 % of the Chrzanowski A, Massiera M (2004) Modeling of deformations during
construction of a large earth dam in the La Grande Complex,
object points and significant radial displacements occurred
Canada. Technical Sciences, Abbrev. Techn. Sc., no 7
on 71 % of the object points. The largest displacements were Chrzanowski A, Chen YQ, Leal J, Murria J, Poplawski T (1991) Use
determined on the upstream part of the crest and 0+230 and of the GPS for ground subsidence measurements in Western
0+460 profiles of the embankment. The maximum horizon- Venezuela oil fields. Land Subsidence (Proceedings of the
Fourth International Symposium on Land Subsidence, May
tal displacement was determined as 14.12 cm with a radial
1991). IAHS Publ. no. 200.
component of 14.08 cm for 4.5 years. Clements RP (1984) Post-construction deformation of rockfill dams. J
Although it has been observed that the monthly average Geotech Eng 110(7):821–840
radial velocity shows a decreasing trend with time, there is a DeLoach SP (1989) Continuous deformation monitoring with GPS. J
Survey Eng 115(1):93–110
partial increase at the last period with respect to reservoir Department of the Interior, Bureau of Reclamation and International
water level. The profiles have inverse curvature while com- Water Power and Dam Construction (2007) World’s largest dams.
pared with crest arch. Cross-sectional movements have larg- http://www.infoplease.com/ipa. 11 April 2011
er values towards to the upstream. As a result, it is difficult Einfopedia (2012) Top ten highest dams in the world. http://www.einf
opedia.com/top-ten-highest-dams-in-the-world.php. 13 Nov 2012
to conclude that the radial displacement on the crest depends
Electrowatt Engineering Ltd; Dolsar Engineering Ltd (2004) Atatürk
exactly to the reservoir water level. Dam and hydroelectric power plant geodetic dam monitoring May
Both conventional and GPS measurement techniques 2004. Technical Report. Electrowatt Engineering Ltd., Zurich.
were applied in this study and it was found that the results Dolsar Engineering Ltd, Ankara
Gikas V, Sakellario M (2008) Settlement analysis of the Mornos earth
obtained from these two techniques were compatible to each
dam (Greece): evidence from numerical modeling and geodetic
other since the positional accuracy was better than ±1 cm. monitoring. J Eng Struct 30(11):3074–3081
This accuracy is appropriate to monitor deformations that Hudnut KW, Behr JA (1998) Continuous GPS monitoring of structural
occurred on the rock fill dams and their surrounding areas deformation at Pacoima Dam, California. Seismol Soc Am 69
(4):299–308
such as Ataturk Dam. Kalkan Y, Alkan RM, Bilgi S (2009) Deformation monitoring of
Ataturk Dam. Proceedings of the International Emergency
Acknowledgments The author would like to thank the Turkish Gen- Management Society. 16th TIEMS Annual Conference-TIEMS,
eral Directorate of State Hydrologic Works for supporting the project June 09–11 Istanbul, Turkey, pp 300–310.
titled “Monitoring deformations of Ataturk Dam using Geodetic Meth- Kalkan Y, Alkan RM, Bilgi S (2010) Deformation monitoring studies
ods,” staff working on the Ataturk Dam for their contribution during at Ataturk Dam. Proceedings on XXIV FIG International Congress,
field survey, and researchers of Istanbul Technical University, Geo- Sydney, Australia.
matics Engineering Department for assisting the field work. The author Kim YS, Kim BT (2008) Prediction of relative crest settlement of
also would like to thank Professor Haluk Özener and anonymous concrete-faces rockfill dam analyzed using an artificial neural
reviewers whose comments mostly inspire the quality of this paper. network mode. J Comput Geotech 35(3):313–322
Malla S, Wieland M, Straubhaar R (2007) Assessment of long-term
deformations of Ataturk Dam. 1st National Symposium and
Exposition on Dam Safety, Ankara, Turkey.
References Özkan ŞÖ, Yılmaz Ş (1999) Türkiye’deki barajlar ve hidroelektrik
santraller. DSİ Genel Müdürlüğü, Ankara
Sedlak V, Jecny M (2004) Deformation measurements on bulk dam of
Aguilera DG, Lahoz JG, Serrano J (2007) First experiences with the waterwork in East Slovakia. Geology, Ecology, Mining Service
deformation analyses of large dam combining laser scanning and 50(2):1–10
high accuracy surveying. XXI International CIPA Symposium, Taşçı L (2008) Dam deformation measurements with GPS. Geodes
01–06 October, Athens, Greece. Cartogr 34(4):116–121
Aguilera DG, Lahoz JG, Sanchez J (2008) A new approach for struc- Taşçı L (2010) Analysis of dam deformation measurements with
tural monitoring of large dams with a three-dimensional laser the robust and non-robust methods. Sci Res Essays 5
scanner. Sensors 8:5866–5883 (14):1770–1779
Bilgi S, Kalkan Y, Yalın SD, Erden T, Yavaşoğlu H (2006) Geodetic Turkish General Directorate of State Hydrologic Works (2005) http://
monitoring in the biggest dam of Turkey: Ataturk Dam. www2.dsi.gov.tr/english/. 12 April 2011
International Symposium on Modern Technologies, Education USACE (2002) Engineering and design structural deformation survey-
and Professional Practice in Geodesy and Related Disciplines, ing (EM 1110-2-1009). Department of the Army US Army Corps
November 9–10, Sofia, Bulgaria, 94–102. of Engineers, Washington, DC 20314-1000.
Bureau of Reclamation (2012) What is the biggest dam in the world? Vladimír S, Miloš J (2004) Deformation measurements on bulk dam of
http://www.usbr.gov/lc/hooverdam/History/essays/biggest.html, waterwork in East Slovakia. Vol 50, No. 2, p. 1–10, ISSN 0474-
13 Nov 2012 8476