Content uploaded by Halil Kumsar
Author content
All content in this area was uploaded by Halil Kumsar on Jun 23, 2016
Content may be subject to copyright.
Content uploaded by Halil Kumsar
Author content
All content in this area was uploaded by Halil Kumsar on Jun 23, 2016
Content may be subject to copyright.
Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag 99
Engineering geological
characteristics of the 1998
Adana-Ceyhan earthquake,
with particular emphasis on
liquefaction phenomena
and the role of soil
behaviour
R. Ulusay 7 Ö. Aydan 7 H. Kumsar 7 H. Sönmez
Received: 4 January 2000 7 Accepted: 28 March 2000
R. Ulusay (Y) 7 H. Sönmez
Hacettepe University, Department of Geological Engineering,
06532 Ankara, Turkey
e-mail: resat6hacettepe.edu.tr
Fax: c90-312-2352542
Ö. Aydan
Tokai University, Department of Marine Civil Engineering,
Orido, 3–20–1, Shimizu, Shizuoka-Ken, 424, Japan
H. Kumsar
Pamukkale University, Department of Geological Engineering,
Denizli, Turkey
Abstract The Adana-Ceyhan earthquake (M
s
p6.2)
occurred in the southern part of Turkey on 27 June
1998 and resulted in the loss of 145 lives and exten-
sive damage to buildings in Ceyhan town and the
settlement areas in its vicinity. Soil liquefaction,
ground failure due to lateral spreading and rock falls
occurred. The area of Adana is characterised by a
large alluvial basin with a delta shape. Most of the
basin is filled with Quaternary recent Holocene
deposits. The recent rapid deposition of sediments
and the very shallow groundwater table throughout
the basin create conditions conducive to liquefac-
tion. The results of a preliminary investigation of
soil liquefaction caused by the earthquake and lique-
faction assessments based on field performance data
are presented together with evaluations concerning
the likely contribution of the soils to the damage
sustained by buildings. The results of the liquefac-
tion susceptibility analysis indicated that the data
from the liquefied sites were within the empirical
bounds suggested by the field-performance evalua-
tion method. It was also shown that shallow sand
layers should have liquefied and the surface disrup-
tion observed on the site could be predicted by the
bounds used for the relationships between the thick-
ness of liquefiable sediments and the overlying non-
liquefiable soil. Site-response analyses based on
acceleration response spectra from the actual earth-
quake’s strong motion records revealed that soil
behaviour was one of the most significant factors in
the damage to buildings caused by the earthquake.
Résumé Le tremblement de terre de Adana-Ceyhan
survenu au sud de la Turquie le 27 juin 1998 á cause
la perte de 145 vies humaines, des dommages éten-
dus aux édifices dans la district de Ceyhan et la
subsidence des alentours. La liquéfaction du sol, des
mouvements latéraux de terrain et des choutes
rocheuse sont aussi survenus dans la district. Dans
cet article, les resultats d’une étude préliminaire de
la liquéfaction du sol causée par le tremblement de
terre, l’évaluation des liquéfaction basées sur les
données terrain expérimentale ainsi que la contribu-
tion possible du sol sur les dommages causées aux
édifices sont présentés. La region de Çukurova est un
delta caracterisé par une grande plaine alluviale. Le
pluspart de la region á été remblayée par des depots
récent d’Holocène de l’aire Quaternaire. L’accumula-
tion rapide des sédiments récents ainsi que la faible
profoundeur de la nappe superfacielle dans le bassin
R. Ulusay et al.
100 Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag
a conduit à des conditions favorables à l’apparition
de liquéfaction du sol. Les resultats des analyses de
susceptibilité de liquéfaction ont montré que les
données provenant des sites du sol liquéfiés sont
dans les limites empiriques de la methode d’évalua-
tion des performances de terrain. Il a aussi été
montré que des couches de sable peu profondes ont
pu étre liquéfiées et que les données provenant des
sites avec ruptures de surface ont été predités dans
les limités utilisées par les relations entre épaisseur
des sediments liquéfiables et l’épaisseur du sol non-
liquéfiable des couches supéreures. Les analyses de la
response des sites basées ou les spectres de response
des accelérations provenant des enregirstements des
actuels tremblement de terre, ont revelé que le
comportement du sol a été un au des facteurs
majeurs sur les dommages des édifices causés par le
tremblement de terre.
Key words Çukurova basin 7 Earthquake 7 Lateral
spreading 7 Liquefaction 7 Site effects
Mots clés Bassin de Çukurova 7 Tremblement de
terre 7 Etendre lateralement 7 Liquéfaction 7 Les
effects d’emplacement
Introduction
The province of Adana is the fourth largest in Turkey with
a population of about 2.3 million. It includes 13 townships,
notably Adana, Ceyhan and Yüreg˘ir. At 16: 55 :53 (local
time) on 27 June 1998, a major earthquake of magnitude
M
s
p6.2 (USGS) and M
s
p6.3 (Kandilli Observatory,
Turkey) struck the Adana Province. Instrumentation indi-
cated epicentre coordinates of 36.945 N and 35.305 E and
macroseismic coordinates of 36.94 and 35.55 E (Fig. 1). It
was reported that the earthquake was felt in the city of
Ankara – 400 km away from the epicentral area of the
earthquake – and as far south as Cyprus, Syria, Israel and
Jordan. Its duration was about 20 s.
Official estimates place the death toll at about 145 and the
number of injured at about 1500. The areas severely
shaken by the earthquake covered an area of approxi-
mately 150 km radius, but the damage occurred within a
30-km radius of the epicentre. Inventory studies by the
General Directorate of Disaster Affairs of Turkey showed
that 1388 buildings suffered heavy damage, 18,612
sustained moderate damage and 43,644 buildings escaped
the earthquake with only slight damage. A total of seven
apartment buildings collapsed in Ceyhan town centre
alone, with the loss of 76 lives. As it was day time, a large
proportion of the residents were not in the towns and
villages or the death toll could have been much higher. The
estimated economic loss due to structural damage alone is
about 870 million US dollars.
Fig. 1
Geographical location of the Adana province, epicentre of the
earthquake and locations of boreholes selected for liquefaction
assessments
The Earthquake Research Department (ERD) estimated
that the epicentre of the main shock was located about
30 km south-east of Adana and 32 km away from the
centre of Ceyhan town, as shown in Fig. 1. The focal depth
for this event is reported by the ERD as 23 km. While no
foreshock was felt, 220 aftershocks with magnitudes
varying between 2.2 and 5.1 had been recorded by July
1998.
Clear liquefaction traces along the banks of the Ceyhan
River were observed west of Ceyhan town, extending to
distances of up to 50 km from the epicentre. Several struc-
tures of liquefied subsoil, such as isolated sand volcanoes
or boils, grouped sand boils, vent-fractures or cracks and
lateral spreading were evident, while lateral spreading
ground failures developed on the banks of the Ceyhan
River. On the steep slopes along the fault zone activated by
the earthquake, major phenomena associated with the
intensity of this dynamic event were also manifested. The
1998 earthquake, Adana-Ceyhan, Turkey
Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag 101
secondary effects of particular importance were the rock
falls observed on some hills in the southern part of the
province, while in the developed areas, fissures and ground
displacements disrupted fields, roadways, canals, buildings
and other constructed works. It is also noted that the soil-
structure interaction of the rather stiff buildings may have
contributed to lengthening of the buildings’ periods to
coincide with the dominant periods of the site.
Soil liquefaction and the effect of the geology on the
damage caused by the 1998 Adana-Ceyhan earthquake
were investigated by the authors. In this paper, the engi-
neering geological characteristics of the earthquake, its
geological and seismological aspects, observations of the
soil liquefaction phenomena, the distribution of the areas
of liquefaction, the nature of the liquefaction and the engi-
neering properties of the soils are discussed. The liquefac-
tion characteristics and the behaviour of the soils are then
examined in respect of their influence on the damage
sustained by buildings. Geotechnical data, including bore-
hole records, laboratory test results from the liquefied soil
samples taken from both sand boils and boreholes were
taken into account in the liquefaction analyses. The lique-
faction assessments were performed by using the field
performance data obtained from standard penetration
tests (SPT). In addition, in order to assess the influence of
the soil behaviour on the damage to buildings caused by
the earthquake, the natural periods of soil columns
selected at particular areas were computed and compared
with those of buildings. For the purpose, a short-cut proce-
dure based on the shear wave distribution of soil layers,
soil thickness and the number of stories of the buildings
was employed.
The field studies were conducted during June 1998, 2 weeks
after the earthquake, and related additional information
was collected during August 1998.
Geological context
The south-east Mediterranean region where the areas
affected by the recent earthquake are located consists of
three basins separated by two main tectonic zones striking
NE–SW (Fig. 2a). The basement rocks of Pre-Miocene age
vary between the three basins as they are located in
different tectonic zones (Kozlu 1987): the I
˙skenderun Basin
being situated in the Amanos Mountains and in the
vicinity of I
˙skenderun Bay, the Adana Miocene Basin in the
southern part of the East Taurus Belt, and between these,
the Misis-Andırın Basin (Fig. 2a).
Geological setting
The area between the Adana and Misis-Andırın Basins,
where the recent earthquake was severely felt, is geographi-
cally known as the Çukurova Basin and is characterised by
a very large alluvial basin with a delta shape which extends
more than 100 km east–west and approximately 70 km
north–south. Most of this basin is filled with Quaternary
recent (Holocene) deposits (Fig. 2b). According to Kozlu
(1987), the geological units of the Çukurova Basin belong
to the Upper Cretaceous, Oligo-Miocene, Miocene, Plio-
cene and Quaternary periods. The rock units of the Upper
Cretaceous are found in the south-east of the basin and
show a volcano-sedimentary stratigraphy consisting of
tuffs, sandstones, marls, agglomerates and huge blocks of
limestones.
The Oligo-Miocene series trend NE–SW, almost parallel to
the Mediterranean Sea coast (Fig. 2b), and are character-
ised by olistostromes. Conglomerates and sandstones from
the Miocene outcrop in the southern and northern parts of
the basin and are known as the Güvenç Formation (Yetis¸
1988). The rock units are generally moderately bedded and
rarely cross-bedded. The Handere Formation described by
Schmidt (1961) represents the Pliocene period and covers
large areas in the northern part of Adana city. This is
mainly composed of sandstone, siltstone, marl and
mudstone and has a thickness of about 700 m (Yetis¸
1988).
In the Çukurova Basin the Quaternary deposits include
alluvial soils and travertines (caliche). Outcrops of the
travertine are visible in the northern part of the basin
between Adana and Ceyhan and also around the Upper
Cretaceous aged blocks in the Cebelinur mountain in the
south-east (Fig. 2b). The travertine and terrace deposits in
the basin have an inclination ranging between 2 and 47
towards south. Their thickness increases towards the
south, to a maximum of 30 m. This unit can be divided
into two layers: a hard caliche with a thickness ranging
between 2 and 6 m at the top and a weak and soft caliche
layer with carbonate gravels at the bottom.
The Çukurova Basin is a broad, flat, fault-bounded plain
traversed by two main rivers, the Ceyhan River in the east
and Seyhan River in the west (Figs. 1 and 2b). When the
Ceyhan River enters the east margin of the basin, the
gradient decreases from a rather steep descent in the
mountains to the almost flat surface of the basin floor. This
sudden change of gradient causes rapid deposition of sedi-
ment. The deposition and redistribution of the Quaternary
sediments carried into the basin by these rivers has led to
the accumulation of thick unconsolidated alluvial plain
deposits composed of intercalated gravel, sand, silt and
clay layers. The recent alluvium is restricted to the beds of
the Ceyhan and Seyhan Rivers. Because of the rapid and
dynamic sedimentation processes, the deposits vary from
well to poorly graded and are generally loosely compacted.
Boreholes undertaken by the General Directorate of State
Hydraulic Works of Turkey (DSI) and private companies
indicate that the thickness of unconsolidated materials
underlying the plain ranges from 100 to 300 m. Figure 3
shows a N–S geological cross section between Ceyhan and
Mercimekköy along which liquefaction fissures and sand
boils were observed.
The alluvial sequence thickens towards the northern part
of the basin, reaching up to 320 m at Mercimekköy and
R. Ulusay et al.
102 Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag
Fig. 2
a Tectonic map of the Adana-
Misis-Andırın-I
˙skenderun
region showing the main
geological basins. (After
Kozlu 1987). b Simplified
geological map of Çukurova
Basin. (Adapted from Kozlu
1987, General Directorate of
Mineral Research and Explo-
ration of Turkey’s 1: 100,000
scale map and Nurlu 1998)
then gradually becoming thinner. The thickness is about
170 m beneath the town of Ceyhan (DSI 1984). The depth
of the clayey surface varies between 1 and 6 m. In Adana,
for example, it is reported to be approximately 1 to 3 m.
Below the clay, the alluvium is generally loose and gravelly
or dense and hard with pockets of sand/clayey sand. Along
a N–S direction the alluvial deposits mainly consist of
clayey and silty material, although sandy and gravelly
layers are found at shallow depths. Towards the east,
between Ceyhan and Osmaniye (Fig. 1), the unconsolidated
materials are coarser grained compared to those on the
Ceyhan plain. The liquefied sandy layers are well seen near
the Ceyhan River. On the basis of the authors’ observations
along the banks of the Ceyhan River and trench studies
undertaken by the ERD (Demirtas¸ 1998) in Yakapınar and
Ceyhan, it is considered that beneath the silty and clayey
horizons, the depth of the mainly fine-grained liquefied
sand layers ranges from 4 to 7 m.
1998 earthquake, Adana-Ceyhan, Turkey
Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag 103
Fig. 3
N–S geological cross section
between Ceyhan and Merci-
mekköy (adopted from DSI
1984)
Tectonics
The area of the Çukurova Basin and its immediate
surrounds has suffered and is still undergoing compression
in a N–S direction and extension in an E–W direction, due
to internal deformation of the Anatolian Block which is
being compressed by the African and Arabian plates to the
south moving northwards against the stationary Euro-
Asian plate in the north (Fig. 4). The Anatolian plate,
bounded by the North Anatolian Fault (NAF) and the East
Anatolian Fault (EAF), is pushed westward against the
Agean plate as a result of the relative movement of the
Arabian plate with respect to the African plate. As a conse-
quence, the Çukurova Basin is bounded by the Ecemis¸ fault
in the west and Yumurtalık-Karatas¸ fault in the east
(Fig. 2a).
The Ecemis¸ fault is a lateral strike-slip fault, the strike
being in a NE–SW direction. Although this is considered
an active fault, there are no written records of a large event
along it . The strike of the Yumurtalık-Karatas¸ fault is one
of the southern splays of the East Anatolian Fault zone
extending towards Cyprus and has a strike almost parallel
to the Ecemis¸ fault. Between these two faults is the Çiçekli-
Göksu fault zone. The left-lateral fault which caused the
1998 earthquake is known as the Misis-Ceyhan fault and
presumed to be a segment of the Çiçekli-Göksu fault.
Kozlu (1987) suggests that this fault was initially the front
of the Yumurtalık thrust and developed during the Lower
Miocene. It was subsequently transformed into a left-
lateral strike-slip fault in the Upper Eocene and Lower
Oligocene. All these fault segments are known to be left-
lateral strike slip faults.
Hydrogeological conditions
The longest river in the region is the Ceyhan River (see
Fig. 1) which originates from the mountains in the north-
east of the basin and carries substantial flows of water and
sediments into the basin throughout the year. Some of the
riverine water infiltrates into the basin through the distri-
butary channel systems, the remainder flowing overland
towards its delta located on the coast of I
˙skenderun Bay.
From the available records of the boreholes drilled at
different times and in different locations throughout the
basin, it is evident that the groundwater table is generally
very shallow. Figure 5 shows the monthly average ground-
water level and average precipitation at a well in the north-
east of the basin (Osmaniye, east of Ceyhan) between 1970
and 1972 (DSI 1975). Although this location is not one
where extensive liquefaction was observed, it is a typical
example indicating the variation in groundwater level in
the Çukurova Basin. The groundwater level is closely asso-
ciated with the amount of precipitation and may be quite
high when the monthly precipitation is high. The records
indicate that in years with high precipitation, the highest
groundwater level was observed in June. The groundwater
level generally fluctuates by between 2 and 6 m near the
R. Ulusay et al.
104 Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag
Fig. 4
Major tectonic elements of
Turkey. (After Barka and
Kadinsky-Cade 1988)
Fig. 5
Monthly averages of groundwater level and average precipitation
at a well in Mamure-Osmaniye, to the east of the Çukurova Basin
(data from DSI 1975)
Ceyhan River, but because of the El Nin˜o phenomenon, the
precipitation in 1998 was quite high and, particularly in the
epicentral area, was likely to have been within 1 to 3 m of
the surface at the time of the June 1998 earthquake.
On the basis of the latest evaluation of the basin by the DSI
(1984), the depth of static water level generally ranges from
between 2 and 5 m below the surface (Fig. 6). These figures
were also confirmed by the records of the boreholes drilled
for different purposes in different parts of the basin. Both
north of Mercimekköy in a large NW–SW trending area
with an elliptical shape and in another small area (Fig. 6),
artesian conditions were reported by the DSI (1984).
Shallow groundwater levels and artesian conditions were
also confirmed by boreholes drilled more recently by the
Turkish Highway General Directorate (TCK) (Çetin 1995).
The direction of groundwater flow through the aquifer in
the alluvium is towards the Ceyhan River. In general,
impervious clayey and silty layers underlie the sandy-grav-
elly layers and/or semipermeable silty clay at shallow
depths. Both the recent rapid deposition of the sediments
and the high groundwater levels contributed to the crea-
tion of conditions favourable to the occurrence of liquefac-
tion.
Faulting and seismic
characteristics of the earthquake
Faulting
The earthquake parameters of the main shock on 27 June
1998 provided by the ERD indicated a location slightly to
the south of Abdiog˘lu village, halfway between Adana and
Ceyhan (Fig. 1). The fault plane postulated by the ERD is
consistent with a strike slip earthquake along a SE-dipping
left-lateral slip fault with a NE–SW strike direction and a
normal component (Fig. 7a). The geometrical position of
the fault plane established by the authors using the hypo-
centre data of the main shock and aftershocks is very
similar to the fault planes postulated by various workers
(Fig. 7b). On the basis of the distribution pattern of the
ground ruptures, aftershocks and fault planes associated
with the main shock, it is concluded that this earthquake
may have originated from the reactivation of a fault plane
related to the known Misis-Ceyhan fault line
Seismicity
Adana and Ceyhan are in the second seismic zone,
according to the seismic zoning map of Turkey prepared
by the ERD. The plot of earthquake epicentres in the basin
from 1900 to 1998 (Fig. 7c) indicates that earthquakes are
generally associated with known active faults existing in
the area. Although the region is known to be seismically
active, because of the short length of the faults in the area,
large earthquakes (M17) are not historically known or
expected. Using these data of earthquakes in the region for
a period between 1881 and 1986, Gençog˘lu et al. (1990)
suggested the following relation for the provincial earth-
quake occurrence for a period 133 years:
logNp3.42P0.51 M (1)
1998 earthquake, Adana-Ceyhan, Turkey
Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag 105
Fig. 6
Static groundwater level map
of Ceyhan and its vicinity.
(DSI 1984)
Fig. 7
a Fault plane associated with
the earthquake as postulated
by the ERD. b Spatial view of
fault plane predicted by the
authors from distribution of
aftershocks. c Distribution of
epicentres of earthquakes
(M
s
64.0) in the Çukurova
Basin and its immediate
vicinity. (After Demirtas¸
1998)
An earthquake with a magnitude of 5.7 occurred at Misis
(see Fig. 7c) on 20 March 1945. Using the above formula
for an earthquake with a Richter magnitude of 6, the recur-
rence interval for a similar earthquake would be 57.9 years.
Considering the earthquakes of 20 March 1945 and 27 June
1998, it would appear that the above formula holds for the
earthquake recurrence in the Adana-Ceyhan province.
The Çukurova Basin has been well instrumented with
several seismic networks set up independent of each other.
Two magnitude Mp6 earthquakes have been recorded,
R. Ulusay et al.
106 Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag
Fig. 8
NS, EW and UD accelerations of the main shock recorded by the
strong motion station located in the local branch building of the
Agricultural Ministry in Ceyhan. (Earthquake Research Depart-
ment 1998)
one in the year 1908 and the other in 1945, almost in the
same location as that of the 27 June 1998 earthquake. The
hypocentre location appears to be a seismic gap in the
region, with almost no seismic activity recorded since 1947
and there was no foreshock to warn people. The main
earthquake shock was recorded by the ERD in the local
branch building of the Agricultural Ministry in Ceyhan
town. According to the data released by the ERD, 104 after-
shocks occurred during the first 2 days following the main
earthquake and 224 aftershocks with magnitudes varying
between 2.2 and 5.0 had been recorded by 17 July 1998.
Assessment of ground motions
The strength of the accelerations recorded at the Ceyhan
Station, 32 km away from the epicentre, were 0.223, 0.273
and 0.086 g in the SN, EW and vertical (UD) directions,
respectively (Fig. 8). The UD component of this earthquake
was also quite high compared with other earthquakes in
Turkey and it is likely that this is characteristic of inland
earthquakes. Aydan et al. (1996) developed a database
system for the seismic characteristics of Turkish earth-
quakes and postulated several empirical relations between
different seismic parameters. The peak ground accelera-
tions measured at other stations during the main shock
enabled the authors to determine the attenuation charac-
teristics of the ground motion. The peak ground accelera-
tion values of the June 1998 earthquake are plotted in
Fig. 9 together with empirical relations proposed by Aydan
et al. (1996), Joyner and Boore (1981), Fukushima et al.
(1988) and the ERD. It can be seen that the attenuation
data are in good agreement only with the empirical rela-
tions developed for Turkish earthquakes by Aydan et al.
(1996). This strong relationship indicates the validity of
using the solid attenuation curve shown in Fig. 9 when
considering Turkish earthquakes.
Engineering geological
characteristics of the earthquake
Lateral spreading and rock falls
Liquefaction-induced lateral spreading of ground failures
was observed along the banks of the Ceyhan River, particu-
larly in Mercimekköy, Abdiog˘lu and Sirkeli on the west
bank of the Ceyhan River. Here the ground is relatively flat
(0.5–1.5%) and movement appears to have been in a north-
easterly direction (Fig. 10).
Close examination of the soil slope failures implied that
they were likely to be a consequence of ground failure due
to liquefaction. For example, a part of a lemon tree planta-
tion, located on the riverbank near Abdiog˘lu village, slid
towards the river and a small pond behind the failed mass
was still visible even 2 weeks after the earthquake
(Fig. 11a). The cracks shown in Fig. 11a were about 10 m
long, through which very thin sandy grains of the alluvium
1998 earthquake, Adana-Ceyhan, Turkey
Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag 107
Fig. 9
Comparison of attenuation data of the Adana-Ceyhan earthquake
with empirical relations proposed by different investigators
were pushed up in the plantation. A vertical subsidence of
2 to 3 m and a horizontal displacement of about 1 to 2 m
towards the river were observed in this vicinity. Another
typical example of lateral spreading, near Mercimekköy,
resulted in a total vertical subsidence of 2.5 m and very
long cracks parallel to the river bank (Fig. 11b). The width
and depth of these cracks ranged from 0.1 to 1 m and 0.5
to 2 m respectively.
Other slope failures were observed on the side of a lined
canal between Misis and Abdiog˘lu, and on the southern
lined bank of a stream near Sirkeli, very close to Ceyhan
(Fig. 11c). The failure direction at both locations was
N20 E, similar to that of the lateral spreads mentioned
above. These ground failures occurred in recent soft allu-
vial deposits and confirm the general experience that a
very specific soil behaviour can be expected particularly in
fine-grained recent alluvial deposits, such as river banks.
Fortunately, as there were no structures of great impor-
tance in the area, the lateral spreading did not cause any
major damage. Indeed, with the exception of the above
examples, no severe damage to structures due to lateral
spreading was observed.
Minor rock falls (maximum 3 to 4 ton) took place on the
south-eastern side of the basin, behind Nacarlı village
where the terrain is mountainous. The location of the rock
falls is shown in Fig. 10. These small-scale falls were asso-
ciated with limestone blocks separating at existing joints in
the rock mass.
Soil liquefaction
Clear traces of liquefaction were observed along the banks
of the Ceyhan River west of Ceyhan town. This phenom-
enon is due to the presence of fine-grained sandy layers in
the alluvial sequence which became liquefied during the
main shock. Several structures typical of liquefied subsoil
were observed, including vent fractures or fissures as much
as several metres wide and isolated sand volcanoes or boils
which ejected water and sediment over areas of up to tens
of square metres.
During the site investigation, the general distribution of
ground fissures and sand boils in the basin was surveyed
(Fig. 10). The longest and widest fissures and the largest
sand boils were concentrated along a 50-km-long north-
east-trending segment of the Ceyhan River. The fissures
and sand boils shown in Fig. 12a,b are typical examples of
the larger liquefaction features which were still clearly
visible 2 weeks after the earthquake. The height and diam-
eter of the sand volcanoes ranged from 100 to 300 mm and
150 to 60 mm respectively. According to local people,
liquefied soil and water rose up to 7 to 8 m above the
surface during the main shock.
No distinct fault scarp was observed, although many en-
echelon-type ground fractures with and without sand boils
were seen near Abdiog˘lu village and Asmalı bridge
(Fig. 12b). All were stepped to the left. In general, the
direction of the eruption fissures and fractures was
N50–70 E and N60–70 W in the southern and northern
parts of Ceyhan town, respectively (Fig. 10). As the course
of the Ceyhan River is parallel to the Misis-Ceyhan fault,
the orientation of eruption fissures generated by liquefac-
tion is generally similar to the course of the Ceyhan River
and the fault seen in Fig. 10.
Observations at the various liquefaction sites indicated
that the soil that erupted from the ground surface was a
dark-grey fine sand with a negligible amount of fines. It
was not possible to determine the depth of the liquefied
layer from the surface observations, but from evidence at a
location near Mercimekköy in the north of Ceyhan, it is
likely this layer lies some 4 to 5 m below the ground
surface, generally at the level of the Ceyhan River. At that
site, traces of the liquefied sand could be seen on the frac-
ture crevasse where ground failure had occurred due to the
lateral spreading (Fig. 10b). This vent fracture was filled by
fine sand. The liquefied sand was generally overlain by a
light-brown non-liquefied stiff alluvial soil layer consisting
of silt and clay (Fig. 12c). However, observations on the
walls of trenches undertaken by the ERD near Abdiog˘lu
village and in Ceyhan town to investigate the shallow stra-
tigraphy beneath the ground failure zone indicated that the
depth of the liquefied sandy layers ranged from 3 to 7 m
from the surface. Demirtas¸ (1998) reported that while the
width of the liquefied sand dykes was 10 to 20 mm imme-
diately below the ground surface, they could reach some
100 mm at 4 m deep.
Although it was anticipated that soil liquefaction would be
widespread along the Ceyhan River, sand boils and other
liquefaction-induced surface features were not evident on
the ground in Ceyhan town. This was supported by the
presence of subvertical sand dykes which did not reach the
surface but terminated at a depth of 3 m below ground
level. This would suggest that the liquefied sand layers are
shallow seated, probably due to the thick, partially satu-
R. Ulusay et al.
108 Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag
Fig. 10
Location of liquefaction sites,
rock falls and slope move-
ment (lateral spreading)
rated, plastic silt and clay layer which capped the silty sand
layer. These less permeable materials would have contrib-
uted to the liquefaction conditions by preventing rapid
upward drainage, except through fissures which may have
developed along the margins of the ground failure zone.
The evidence of foundation disruption or distress beneath
the structures was very limited. However, the effects of
liquefaction on small structures were observed to the
south-west of Ceyhan, in the vicinity of Abdiog˘lu village
where extensive liquefaction occurred. The most specta-
cular examples of damage to structures in the village and
its vicinity were the differential settlement of a water-
storage tower by about 50 mm (Fig. 13a), the uplifting of a
concrete water tank beneath this tower and the tilting of a
fountain in the garden of a primary school (Fig. 13b).
Engineering characteristics of the
liquefied and non-liquefied soils
During field investigations, sampling of “typical” soil was
carried out at a number of locations, including Abdiog˘lu
village, Mercimekköy and Büyükmangıt, which are very
close to the town of Ceyhan. Laboratory testing was under-
taken to determine the physical properties of the liquefied
and non-liquefied soils from the uppermost two zones in
the alluvial sequence. Only disturbed sampling of the
liquefied soil was possible during the site investigations,
the material being taken from the sand boils. Although
originating from the liquefied layer, these materials are
partially filtered and segregated during the eruption
1998 earthquake, Adana-Ceyhan, Turkey
Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag 109
Fig. 11
Views of typical soil slope failures: a lateral spreading in a lemon
tree plantation on the Ceyhan River bank, SW of Ceyhan; b
lateral spreading towards the Ceyhan River near Mercimekli
village (north of Ceyhan); c failure at the southern bank of a
stream near Sirkeli village (SE of Ceyhan)
process. In view of this, the laboratory testing consisted of
only the determination of moisture content, specific
gravity and grading characteristics of the liquefied soil. In
addition, the Atterberg limits of the fine-grained non-
liquefied soil were also determined. The physical charac-
teristics of the two soil types were compared and assessed
from the liquefaction point of view. A summary of sample
characteristics is given in Table 1.
Although the majority of the liquefied soils lost most of
their natural moisture content just after boiling, they were
found to have higher moisture content values than the
samples from the non-liquefied soil. Figure 14a shows that
fine- to medium-sand-sized material ranging between 82
and 93% is dominant in the liquefied soil. The grain size
distribution curves of the samples from the liquefied soil
fell between the well-known upper and lower bounds for
liquefaction and were in good agreement with those from
samples from previous earthquakes in Turkey (Fig. 14b).
None of the samples from the liquefied layers met the
requirements of the Unified Soil Classification System,
where a well-graded sandy soil should have C
u
16 and C
c
between 1 and 3. The testing indicated that these samples
were of poorly graded fine to medium sands with a very
low percentage of fines and fell into the SP soil group. The
D
50
of the soil grains of these sands did not show a scatter
and ranged between 0.11 and 0.3 mm, indicating that the
soil is highly susceptible to liquefaction (Iwasaki 1986).
Both the grain size of the soil and the high groundwater
R. Ulusay et al.
110 Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag
Table 1
Characteristics of the liquefied and non-liquefied soil layers investigated
a
Grain size Atterberg limits (%)
Sand Silt Clay
Location G
s
FMC w
(%)
LL PL PI D
50
(mm)
C
u
C
g
Soil
class
Liquefied sands
1 Abdiog˘lu 2.70 26 63 5 5 1 9.5 N.A. 0.3 4.12 1.51 SP
2a Mercimekköy 2.64 61 25 0 13 1 2.8 N.A. 0.15 4.85 2.04 SP
3a Büyükmangıt 2.68 82 11 0 6 1 16.0 N.A. 0.28 3.88 1.6 SP
4a Bank of Ceyhan River
(near Abdiog˘ lu)
2.70 72 10 0 15 3 4.0 N.A. 0.11 3.2 1.8 SP
Non-liquefied cap soil
2b Mercimekköy 2.57 59 1 0 35 5 3.6 N.D. 0.07 - - -
3d Büyükmangıt 2.44 14 2 0 64 20 10.2 35.7 27 8.7 0.02 - - ML
4b Bank of Ceyhan River
(near Abdiog˘ lu)
2.39 27 3 0 58 12 N.D. 27.4 21 6.4 0.04 - - ML
a
G
s
, Specific gravity; F, fine; M, medium; C, coarse; w, water content; LL, PL, PI, liquid limit, plastic limit, and plastic index, respec-
tively; D
50
, mean grain size; C
u
, coefficient of gradation; N.A., not applicable; N.D., not determined
Fig. 12
a Vent fracture and sand boil near Asmalı bridge. b Liquefaction
in a school yard in Abdiog˘lu village (south of Ceyhan). c Trace of
liquefied soil on a fracture crevasse of a soil slope failure along
the Ceyhan River near Mercimekköy
level in conjunction with the active seismic features of the
region result in conditions favourable to the occurrence of
liquefaction.
Based on the laboratory test results obtained from three
specimens, the uppermost soil overlying the liquefiable
layer is a fine-grained soil and from its grain size would
not be expected to liquefy (Fig. 14c). All the samples had
liquid limits of less than 50%, indicating low plasticity.
These soils classified in the ML group and were defined as
silty clay. The reason that liquefaction did not take place in
this layer could be due to the fine-grained composition of
the soil, its mean grain size, its position with respect to the
groundwater table and the amplitude of the acceleration of
the waves (Aydan et al. 1998).
1998 earthquake, Adana-Ceyhan, Turkey
Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag 111
Fig. 13
a Differential settlement of 50 mm beneath a water-storage tower.
b Tilted fountain in Abdiog˘lu village
Evaluation of liquefaction
susceptibility and ground surface
disruption
Although liquefaction is a major cause of earthquake
damage, little harm occurs unless the liquefied condition
leads to some form of ground surface disturbance or
ground failure. As a consequence, the ability to accurately
predict the potential for ground surface disruption is a
major concern for geotechnologists responsible for the safe
siting of construction works.
Ishihara (1985) published preliminary empirical criteria to
assess the potential for ground surface disruption at lique-
faction sites. These are based on relationships between the
thickness of liquefiable layers beneath a site and the corre-
sponding thickness of the overlying non-liquefiable soil
layer (Fig. 15a). The criteria were also evaluated and
supported by Youd and Garris (1995) using borehole data
from field investigations following several earthquakes. To
test Ishihara’s (1985) proposals, liquefaction assessments
were also carried out in this study using the data available
from the records of selected boreholes drilled at different
locations within the basin. Due to the extent of the alluvial
soils in the basin, only a limited number of boreholes
provided satisfactory input data for the assessment. Prime
consideration was paid to selecting boreholes located as
close as possible to the sites where liquefaction phenomena
were observed. Two of the six boreholes selected were
drilled in Ceyhan town after the earthquake for the
purposes of selecting sites for the construction of new
buildings. The locations of the boreholes are shown in
Fig. 1 and the soil and groundwater conditions through the
boreholes are summarised in Fig. 16. The laboratory-deter-
mined unit weight of the soils ranged between 17 and
19 kN/m
3
. The sand and gravelly sand layers had fines
contents of 0 to 33% and from the depth of the ground-
water level measured in the boreholes were assumed to be
fully saturated.
The procedure based on field performance data put
forward by Seed and DeAlba (1986) is that most widely
used both in Turkey and in most countries to evaluate
liquefaction susceptibility and was employed in this study.
For the Turkish earthquakes (Fig. 9), the attenuation rela-
tionship
a
max
p2.8(e
0.9M
,e
P0.025R
P1) (2)
was employed for the estimation of peak ground accelera-
tion (a
max
), where M
s
and R are the magnitude of the
earthquake and distance from the hypocentre respectively.
Depending on the distance of the selected borehole from
the hypocentre, a
max
values ranging from 0.18 to 0.3 g were
estimated.
On the basis of the information from the borehole records
employed in the assessment of liquefaction susceptibility,
standard penetration tests were also carried out according
to ASTM D-1586 (American Society for Testing Materials
1990) test specifications at 1.5-m intervals in the founda-
tion soils. A standard SPT split-spoon sampler with liner
was used, the sampler being driven into the foundation
soils using the drill rods and the donut-type hammer
raised and dropped by two turns of rope. In this study, the
SPT-N (the number of blows provided from the standard
penetration test) values were corrected for striking energy
during the test, and for the length of rods used and bore-
hole diameter in order to obtain N
60
values for each soil
layer. These were then normalised, (N
1
)
60
to provide a
reference confining pressure using the overburden correc-
tion factor (C
N
) proposed by Tokimatsu and Yoshimi
(1983).
R. Ulusay et al.
112 Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag
Fig. 14
Grain size distribution curves for soil samples from liquefied
layers in the Çukurova Basin (a), from other earthquakes in
Turkey (b), and from the non-liquefied layer near Ceyhan (c)
In order to investigate the liquefaction susceptibility,
geotechnical data and seismic parameters were used in a
computer model “LIQUEFAC” developed by the authors.
The program employs well-known assessment methods
based on both field performance data and laboratory cyclic
shear testing. The output consists of a table of input data
and factors of safety against liquefaction for each layer.
The layers with factors of safety greater than 1.2 and
between 1.0 and 1.2 were predicted to be non-liquefiable
and marginally liquefiable respectively. The program was
written in Visual Basic and can run on any type of IBM
PC-compatible computer.
The results of the liquefaction analysis of the layers
throughout the boreholes are summarised in terms of
factor of safety against liquefaction in Fig. 16. Figure 17
shows the relationship between modified SPT value and
cyclic stress ratio (CSR) together with bounds for liquefac-
tion/non-liquefaction proposed by Seed and DeAlba
(1986). The data for Adana-Ceyhan in Fig. 17 are taken
from the work of Ülker et al. (1998) and the authors’ inves-
tigations. It can be seen that the data for the liquefied soils
generally fall within the range of predicted liquefiable
soils.
It is evident from Fig. 16 that with the exception of a band
in BH 5, most of the sand layers yielded factors of safety
lower than unity and/or between 1 and 1.1, indicating
conditions of liquefaction or marginal liquefaction respec-
tively. Due to the very shallow groundwater levels at these
locations, the relatively short distances from the sampling
locations to the epicentre of the earthquake of 27 June 1998
(generally 30 to 40 km), the low fines content of sand
layers and the very low SPT-N values, it is clear these
layers would be susceptible to liquefaction. This is in
contrast to the sand layer encountered in BH 5, near
Osmaniye city (east of Ceyhan) which would not liquefy,
probably due to both its gravelly nature and its distance
from the epicentre (60 km) compared to other boreholes.
In BH 6, close to BH 5, the gravelly sand layer between 0
and 3.2 m yielded very low factors of safety against lique-
faction. Bearing in mind the very low SPT-N values (Np3
and 4; Fig. 16) obtained from this layer, these results seem
questionable when compared to those in BH5 which
possess similar conditions. It is considered that this is
simply an anomaly, as may be expected in any data set
based on field tests.
During the field investigations, the surface effects were
divided into three categories: (1) no observed surface
effects; (2) sand boils and small ground fissures; and (3)
surface effects generated by lateral spreading. Based on
these observations, the results of liquefaction assessment
analysis through the selected boreholes and the trench
studies by the ERD in the town of Ceyhan and its vicinity,
the bounds published by Ishihara (1985) were tested.
Combinations of different thicknesses of liquefiable and
non-liquefiable surface layers from the analyses are plotted
in Fig. 15b,c.
Figure 15b shows the thickness data from the location of
BH 1 and BH 2 in Ceyhan town with a value of a
max
p0.3 g.
Trench studies by the ERD indicated that sand dykes and
1998 earthquake, Adana-Ceyhan, Turkey
Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag 113
Fig. 15
a Boundary curves proposed by Ishihara (1985) for discrimi-
nating between occurrence and non-occurrence of surface effects
of liquefaction, and thickness of liquefied and non-liquefied soil
layers for various types of surface effects of liquefaction in
Ceyhan and its vicinity for 300-gal bound (b) and for 200-gal
bound (c)
sand boils exist at this location where surface disruption
would be correctly predicted by Ishihara’s (1985) bounds.
Figure 15c shows thickness data from BHs 3 to 6 where no
surface disruption was observed. Data on this plot are
from sites with an estimated a
max
between 0.18 and 0.23 g.
All but one of the data plots shown on this figure agree
well with the bounds proposed by Ishihara (1985). The
data from a gravelly sand layer between 0 and 3.2 m in BH
6 satisfy the conditions of liquefaction and fall above Ishi-
hara’s (1985) bound although no liquefaction-induced
ground surface disruption was observed or reported in
Osmaniye and its vicinity. The anomalous results from BH
6 have been referred to above and it is considered that
such an occasional inconsistency may be expected in any
data set based on field tests.
The ground disturbance phenomena suggested by Ishihara
(1985) are not related only to the layer thickness however.
Other factors influencing ground disturbance include the
inclination of the layer, the load acting on the soil and the
variations in the thickness of the layer . The absence of
data representing these factors should be kept in mind as
they may to some extent limit the validity of the assess-
ments given here. Nevertheless, the comparison of the
liquefiable zones from the study site indicates that the
liquefaction potential of some of the thin sand layers
diminishes where there is a thickness of some 5 m of lique-
faction-resistant soil thickness.
Assessment of role of soil
behaviour on damage to buildings
The influence of the soil behaviour on the damage to
buildings caused by this earthquake was assessed by using
both a short-cut procedure and the strong motion records.
The procedure for calculating the natural period of a soil
column and then comparing it with the natural period of
buildings requires that the shear wave distribution of soil
layers be examined. After the earthquake a preliminary
geophysical study was carried out by the ERD, including
shear wave velocity measurements in Ceyhan and in other
small settlement areas, particularly those located along the
course of the Ceyhan River where the effects of the lique-
faction had been clearly observed. Based on the data from
the unpublished records of the ERD, the shear wave veloci-
ties of the soil columns generally range between 150 and
450 m/s.
R. Ulusay et al.
114 Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag
Fig. 16
Ground conditions in boreholes evaluated for liquefaction assess-
ments and results of liquefaction analysis
For practical computations of the natural period for a
homogeneous soil column, the following equation (Dobry
et al. 1976) is commonly used:
T
s
p4H/V
s
(3)
where T
s
is the natural period (in seconds), H is the thick-
ness (in metres) and V
s
is the shear wave velocity of a soil
1998 earthquake, Adana-Ceyhan, Turkey
Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag 115
Fig. 17
Relationship between modified SPT value N
60
and cyclic stress
ratio (CSR) together with bounds for liquefaction/non-liquefac-
tion proposed by Seed and DeAlba (1986) for soils from the
Adana-Ceyhan region. FC Fines content
Fig. 18
Curves of soil periods estimated for thicknesses of two different
soils
Fig. 19
NS, EW and UD acceleration response spectra of the main shock
(h damping values)
column. Considering both the thickness of the unconsoli-
dated deposits underlying the town of Ceyhan and in the
Çukurova Basin and the measured shear wave velocities,
natural soil periods were calculated for these ranges using
Eq. (3). The plot shown in Fig. 18 indicates soil periods
between 0.9 and 8 s for the range of soil thicknesses
considered. From the information in Fig. 18, assuming
homogeneous soil conditions, the natural period of the soil
underlying a large part of Ceyhan and its immediate
vicinity is between 2.0 and 2.5 s.
The response spectra of accelerations measured at Ceyhan
station during the main shock of 27 June 1998 are given in
Fig. 19 for EW, NS and vertical (UD) components. As can
be seen, the NS spectra indicate peaks at periods of 0.33,
0.45 and 0.67 s. The EW spectra analysis shows peaks at
periods of 0.15, 0.25, 0.35, 0.47, 0.68 and 1.10 s, while the
R. Ulusay et al.
116 Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag
UD spectra yields peaks at periods of 0.06, 0.15, 0.35, 0.47,
0.68 and 1.10 s.
Based on the experimental data reported by Bayülke
(1978), the following relations between the natural period,
T
b
, and the number, N, of storeys are realistic for buildings
in Turkey:
T
b
p0.05 N (non-damaged) (4)
T
b
p0.10 N (damaged) (5)
For a proper earthquake design, the number of storeys to
be built should be such that the natural period of the
building does not match that of the soil column. In this
regard, the most undesirable number of storeys for build-
ings in the study area would be 20 to 25, based on soil
periods of 2.0 to 2.5 s computed using the short-cut proce-
dure [Eq. (3)] and Eq. (5). However, the buildings that
suffered the most severe damage in Ceyhan generally were
4 to 8 storeys high. Based on the soil periods obtained
from response spectra (Fig. 19) and Eq. (5), it would be
anticipated that buildings with between 5 and 11 storeys
would be more severely damaged. The fact that most of the
collapsed or heavily damaged apartment blocks in Ceyhan
were between 4 and 8 storeys high supports the conclu-
sions drawn from response spectra analysis. However,
predictions based on Eq. (3), which uses homogenous soil
conditions, are not consistent with field observations. This
is considered to be due to the fact that shear wave veloci-
ties were probably measured for a soil column of 5 to 10 m
thick near the surface rather than the thickness of 300 m
employed in the short-cut procedure.
Bayülke (1999) reported that following the earthquake, the
ERD measured the natural periods of five damaged build-
ings in Adana, between 9 and 11 storeys high. The results
indicated natural periods of between 0.55 and 0.93 s, corre-
sponding with that of the soil column estimated from the
response spectra analysis. Such a coincidence of predomi-
nant periods of ground motion within the range of Tp0.5
to 0.7 s may explain the fact that the most severe damage
was concentrated in the 5- to 8-storey buildings. As a
consequence, it was concluded that there is a good consist-
ency between the results of the experimental study by the
ERD and those from this study.
The acceleration records taken at various stations in the
area where the 27 June 1998 earthquake occurred also
enabled the authors to perform a preliminary assessment
on soil amplification. The peak ground accelerations
measured at eight stations during the main shock and the
epicentral distance of each station provided by the ERD are
presented here in Table 2. Peak accelerations from the
Ceyhan and Mersin stations are considerably higher than
those from other stations and indicate a clear and signifi-
cant soil amplification. It is also noted that although
Ceyhan and Karatas¸ are at approximately equal distances
from the epicentre, the peak ground acceleration measured
at Ceyhan was about eight times greater than that
measured in Karatas¸. A ratio of about 9 between the peak
ground accelerations from Mersin and Hatay indicates a
similar condition. This preliminary assessment indicates
Table 2
Peak ground accelerations recorded at different stations in Çuku-
rova Basin and around. (Earthquake Research Department
1998)
ERD
a
station Distance from
epicentre (km)
Peak acceleration
(cm/s
2
)
Ceyhan 32 273
Karatas¸36 33
Iskenderun 68 15
Mersin 78 132
Hatay 97 27
Islahiye 108 18
K. Maras¸ 149 8.5
Elbistan 210 5
a
ERD, Earthquake Research Department
Fig. 20
View of collapsed and collided 5- and 6-storey residential apart-
ment blocks in Ceyhan with shops at ground floor
that the very soft soil conditions at Ceyhan resulted in
significant soil amplification.
The above assessments suggest that soil behaviour might
be considered as one of the most significant factors in the
damage to buildings caused by the earthquake. The other
main causes of damage were almost the same as those
identified in previous earthquake events in Turkey, such as
poor workmanship, granulometry of the sand and gravels
of concrete, resonance, design errors, weak floors and the
collision of structures (for details, see Aydan et al. 1998).
Figure 20 provides a typical example of the poor construc-
tion conditions in the town of Ceyhan, showing collapsed,
collided and damaged 5- to 6-storey residential buildings
constructed in close proximity and with shops at ground
floor level. Two of the buildings survived the earthquake
with minor to moderate damage while the third was
completely destroyed. The completely different seismic
response of the buildings may be explained either by the
directional effects of the ground motion or by different
1998 earthquake, Adana-Ceyhan, Turkey
Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag 117
configurations of the masonry-infilled walls in the first
storey.
Conclusions
A moderate earthquake (M
s
p6.2) struck the Adana-
Ceyhan province in the south of Turkey on 27 June 1998,
causing much more damage than would be expected for an
event of this magnitude. Based on an investigation of the
liquefaction phenomena, an assessment of the damage to
structures which could be related to soil behaviour and an
evaluation of seismic data, the following main conclusions
are drawn.
1. The seismic characteristics of this earthquake were
generally consistent with empirical relations developed
for Turkish earthquakes. The main fault plane at the site
was predicted to have a strike in the direction of
NE–SW, which corresponds with the Misis-Ceyhan
fault. Based on the postulated fault plane and the distri-
bution pattern of the ground ruptures, liquefaction
features and aftershocks, it can be concluded that the
earthquake is likely to have originated from the reacti-
vation of this fault.
2. Although the shallow liquefied sand layers are not
continuous throughout the Çukurova Basin, the recent
rapid deposition of sediments and very shallow ground-
water table indicate conditions conducive to liquefac-
tion during even moderate earthquakes. Widespread
liquefaction and lateral spreading of the ground were
observed for a distance of 50 km, particularly along the
banks of the Ceyhan River. The damage in Ceyhan was
concentrated in structures built on alluvial deposits,
particularly when parallel to the meandering Ceyhan
River. Soil slope and embankment failures were asso-
ciated with lateral spreading of ground as a result of
liquefaction. The grain size distribution curves of the
liquefied soil from the side fall between the well-known
upper and lower bounds for liquefaction and are in very
good agreement with those of the previous earthquakes
in Turkey. The results of a liquefaction susceptibility
analysis based on data from six boreholes drilled in
Ceyhan and its surrounds indicated that most of the
sand layers were highly susceptible to liquefaction with
factors of safety lower than unity. The data from lique-
fied sites sampled following the earthquake are within
the empirical bounds suggested by Seed and DeAlba
(1986). Comparison of predicted liquefiable zones and
observed liquefaction-induced ground surface disrup-
tions using the criteria suggested by Ishihara (1985)
revealed that the field data were in good agreement with
the bounds of the criteria and that at depths of about
5 m, the liquefaction potential of thin sand layers
diminished.
3. The assessments based on response spectra analysis and
peak ground accelerations from various stations
revealed that the soil behaviour was one of the most
significant factors in the damage sustained by buildings.
The concurrence of the fundamental frequencies of the
4- to 8-storey buildings and the subsoil resulted in
significant amplification of the building excitation from
resonance. The other likely contributory causes of
damage were similar to those in previous earthquakes
in Turkey, including bad design, lack of reinforcement,
poor workmanship and an inappropriate mix of sand
and gravels in the concrete used in the buildings.
4. One important aspect of the Adana-Ceyhan earthquake
is its relatively long duration of about 20 s. Further-
more, the epicentre was close to Ceyhan town which is
founded on thick alluvial soils prone to liquefaction.
These factors are also considered to have contributed to
the damage suffered.
The main lessons learned from this earthquake are the
important influence of local soil conditions on the damage
to buildings and the necessity for both a proper design and
adequate reinforcement of buildings constructed in earth-
quake-prone areas. Bearing this in mind, the policy
concerning urban-regional planning and development
must be reconsidered, based on the best evaluation of the
engineering, geological and seismic risk parameters of the
geological environment. Buildings must meet the design
requirements of the Codes and be constructed to be earth-
quake resistant. A good design and understanding of
ground conditions may result in reducing the extent of
structural damage.
References
American Society for Testing Materials (1990) Standard
method for penetration test and split-barrel sampling of soils
(D1586–84). In: Annual book of ASTM standards, sect 4, vol
04,09. ASTM, Philadelphia, 1092 pp
Aydan Ö, Sezaki M, Yarar R (1996) The seismic characteristics
of Turkish earthquakes. Proc 11th World Conference on Earth-
quake Engineering, Acapulco, Mexico, pp 1–8
Aydan Ö, Ulusay R, Kumsar H, Sönmez H, Tuncay E (1998)
A site investigation of Adana-Ceyhan earthquake of 27 June,
1998. Report TDV/DR 006–30. Turkish Earthquake Foundation,
Istanbul, 132 pp
Barka AA, Kadinsky-Cade K (1988) Strike-slip fault geometry
in Turkey and its influence on earthquake activity. Tectonics
7: 663–684
Bayülke N (1978) Tug˘la yig˘ma yapıların depremdeki davranıs¸ı.
Deprem Aras¸tırma Enstitüsü Bülteni 6(22): 26–41
Bayülke N (1999) Yapı periyodu nedir, tasarımda yapının hangi
anındaki periyot kullanılmalıdır? I
˙ns¸aat Mühendisleri Odası
Adana s¸ubesi Bülteni 19 : 24–26
Çetin H (1995) Kozan-Ceyhan ayrımı Kadirli yolu ve Kadirli-
Osmaniye yolu köprü temel sondajlari raporu. TCK Raporu, 6
pp
Demirtas¸R (1998) 27 Haziran 1998 Ceyhan-Misis depremi; sıvı-
las¸ma ve hendek çalıs¸ması. Jeoloji Mühendislig˘i 53: 40–57
DSI
˙(1975) Osmaniye Ovasi hidrojeolojik etüt raporu, 39 pp
DSI
˙(1984) Kadirli-Kozan-Ceyhan Ovası hidrojeolojik etüt
raporu, 20 pp
Dobry R, Oweis L, Urzua A (1976) Simplified procedures for
estimating the fundamental period of a soil profile. Bull Seis
Soc Am 66(4):1293–1321
R. Ulusay et al.
118 Bull Eng Geol Env (2000) 59 : 99–118 7 Q Springer-Verlag
Fukushima Y, Tanaka T, Kataoka S (1988) A new attenuation
relationship for peak ground accelerations. In: Proc 9th WCEE,
Tokyo
Gençog˘lu S, I
˙nan I
˙, Güler H (1990) Türkiye’nin deprem tehli-
kesi. Türkiye Jeofizik Mühendisleri Odası Yayinı.
Ishihara K (1985) Stability of natural deposits during earth-
quakes. In: Proc 11th Int Conf on Soil Mech and Found Eng,
vol 1. AA Balkema, Rotterdam, pp 321–376
Inan E, Colakoglu Z, Koc N, Bayulke N, Coruh E (1996)
Catalogue of earthquakes between 1976–1996 with acceleration
records. Earthquake Research Department, General Directorate
of Disaster Affairs of Public Works and Settlement, Ankara,
150 pp
Iwasaki T (1986) Liquefaction and flow during earthquakes.
Geotechnique 43(3) :351–415
Joyner WB, Boore DM (1981) Peak horizontal acceleration and
velocity from strong-motion records including records from
the 1979 Imperial Valley, California Earthquake. Bull Seis Soc
Am 71(6):2011–2038
Kozlu H (1987) Misis-Andırın dolayının stratigrafisi ve yapısal
evrimi. Proc 7th Turkish Petroleum Congress, Ankara, Bildir-
iler Kitabı, pp 104–116
Nurlu M (1998) 27 Haziran 1998 Adana Ceyhan depremi. Afet
I
˙s¸leri Genel Müdürlüg˘ü, Deprem Aras¸tırma Dairesi, 21 pp
Schmidt GC (1961) Stratigraphic nomenclature for the Adana
region petroleum district. Proc 7th Petroleum Administration
Publ Bull 6 :47–63
Seed H, DeAlba P (1986) Use of SPT and CPT tests for evalu-
ating the liquefaction resistance of sands. In: Use of in-situ
tests in geotechnical engineering. ASCE Geotech Spec Publ
6: 281–302
Tokimatsu K, Yoshimi Y (1983) Empirical correlation of soil
liquefaction based on SPT-N value and fines content. Soils
Found 23(4) :56–74
Ülker R, Celep Z, Hasgür Z, Eyidog˘an H, Kaynak B (1998)
27 Haziran (1998) Adana-Ceyhan Depremi hakkında. ITÜ Yapı
ve Deprem Uygulama Aras¸tırma Merkezi, I
˙stanbul
Yetis¸C (1988) Reorganization of the Tertiary stratigraphy in the
Adana basin. Newsl Stratig 20(1): 43–58
Youd TL, Garris CT (1995) Liquefaction-induced ground-
surface disruption. J Geotech Eng 121(11) : 805–809
