2007 NatHaz Ismail-Zadeh Sokolov Bonjer

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

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

net/publication/225680946

Tectonic stress, seismicity, and seismic hazard in the southeastern


Carpathians

Article in Natural Hazards · July 2007


DOI: 10.1007/s11069-006-9074-1

CITATIONS READS

11 564

3 authors, including:

Vladimir Sokolov Klaus-Peter Bonjer


Saudi Geological Survey Karlsruhe Institute of Technology
89 PUBLICATIONS 1,425 CITATIONS 58 PUBLICATIONS 2,220 CITATIONS

SEE PROFILE SEE PROFILE

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

The revised regional ground-motion models for western Saudi Arabia View project

PhD-Thesis View project

All content following this page was uploaded by Vladimir Sokolov on 01 March 2016.

The user has requested enhancement of the downloaded file.


Nat Hazards (2007) 42:493–514
DOI 10.1007/s11069-006-9074-1

ORIGINAL PAPER

Tectonic stress, seismicity, and seismic hazard


in the southeastern Carpathians

Alik Ismail-Zadeh Æ Vladimir Sokolov Æ


Klaus-Peter Bonjer

Received: 21 May 2005 / Accepted: 1 July 2006 / Published online: 30 January 2007
 Springer Science+Business Media B.V. 2007

Abstract Intermediate-depth earthquakes in the Vrancea region occur in response


to stress generation due to descending lithosphere beneath the southeastern
Carpathians. In this article, tectonic stress and seismicity are analyzed in the region
on the basis of a vast body of observations. We show a correlation between the
location of intermediate-depth earthquakes and the predicted localization of maxi-
mum shear stress in the lithosphere. A probabilistic seismic hazard assessment
(PSHA) for the region is presented in terms of various ground motion parameters on
the utilization of Fourier amplitude spectra used in engineering practice and risk
assessment (peak ground acceleration, response spectra amplitude, and seismic
intensity). We review the PSHA carried out in the region, and present new PSHA
results for the eastern and southern parts of Romania. Our seismic hazard assess-
ment is based on the information about the features of earthquake ground motion
excitation, seismic wave propagation (attenuation), and site effect in the region.
Spectral models and characteristics of site-response on earthquake ground motions
are obtained from the regional ground motion data including several hundred
records of small and large earthquakes. Results of the probabilistic seismic hazard
assessment are consistent with the features of observed earthquake effects in the
southeastern Carpathians and show that geological factors play an important part in
the distribution of the earthquake ground motion parameters.

A. Ismail-Zadeh (&) Æ V. Sokolov Æ K.-P. Bonjer


Geophysikalisches Institut, Universität Karlsruhe,
Hertzstr. 16, Karlsruhe 76187, Germany
e-mail: alik.ismail-zadeh@gpi.uni-karlsruhe.de

A. Ismail-Zadeh
Institut de Physique du Globe de Paris,
4 Place Jussieu, 75252 Paris, France

A. Ismail-Zadeh
International Institute of Earthquake Prediction Theory and Mathematical Geophysics,
Russian Academy of Sciences, Warshavskoye shosse 79-2,
Moscow 113556, Russia
123
494 Nat Hazards (2007) 42:493–514

Keywords Intermediate-depth earthquake Æ Vrancea Æ Seismic intensity Æ


Peak ground acceleration Æ Probabilistic seismic hazard assessment

Abbreviations
CALIXTO Carpathian Arc Lithospheric X-Tomography
FAS Fourier amplitude spectra
HVSR Horizontal-to-vertical Fourier spectral ratio
MSK Intensity scale
PGA Peak ground acceleration
PSHA Probabilistic seismic hazard assessment
RSA Response spectra amplitude
VHR Very hard rock

1 Introduction

Repeated large intermediate-depth earthquakes in the southeastern (SE-) Carpa-


thians (the Vrancea region) cause destruction in Bucharest (Romania) and shake
central and eastern European cities several hundred kilometers away from the
hypocenters of the events. The earthquake-prone Vrancea region (Fig. 1) is boun-
ded to the north and northeast by the Eastern European platform, to the east by the
Scythian platform, to the south-east by the Dobrogea orogen, to the south and south-
west by the Moesian platform, and to the north-west by the Transylvanian basin. The
epicenters of the mantle earthquakes in the Vrancea region are concentrated within
a very small area. The projection of the foci on a NW–SE vertical plane across the
bend of the Eastern Carpathians (section AB in Fig. 1) shows a seismogenic volume
about 100 km (deep) · 70 km · 30 km, and extending to a depth of about 170 km.
Beyond this depth the seismicity ends suddenly: one seismic event at 220 km depth
represents an exception (Oncescu and Bonjer 1997). According to the historical
catalog of Vrancea events (Radu 1979; 1991), large intermediate-depth shocks with
magnitudes MW > 6.5 occur three to five times per century. In the 20th century, large
events at depths d of 70–170 km occurred in 10.11.1940 (moment magnitude MW = 7.7,
d = 160 km), in 04.03.1977 (MW = 7.5, d = 100 km), in 30.08.1986 (MW = 7.2,
d = 140 km), and in 30.05.1990 (MW = 6.9, d = 80 km) (e.g., Oncescu and Bonjer
1997). According to Sandi (2001) 1,570 people died, 11,300 were injured and 32,500
residential and 763 industrial units were destroyed or seriously damaged as results of
the 1977 Vrancea earthquake.
According to modern concepts on the regional geodynamics, Vrancea interme-
diate-depth earthquakes occur in response to stress generation at these depths due to
descending lithospheric slab (McKenzie 1972; Fuchs et al. 1979; Oncescu 1984;
Linzer 1996; Pana and Erdmer 1996; Girbacea and Frisch 1998; Pana and Morris
1999; Ismail-Zadeh et al. 1999, 2005a, b, 2000; Sperner et al. 2005, 2001; Ismail-
Zadeh, 2003). The primary models of geodynamics, stress generation, earthquake
occurrences, and strong ground motions caused by earthquakes are important inputs
for seismic hazard analysis. Seismic hazard assessment in terms of engineering
parameters of strong ground motion (namely, peak ground acceleration PGA,
response spectra amplitude RSA, and seismic intensity) is based on the information

123
Nat Hazards (2007) 42:493–514 495

Fig. 1 Observed seismicity in Romania for 1990–2002 with magnitude MW ‡ 3 and location of
CALIXTO (triangles) and K2 (circles) stations. (a) Epicenters of Vrancea earthquakes determined
by the joint hypocenter method. The background is the topography; the two bold lines show the
location of the refraction seismic profiles VRANCEA99 (N–S) and VRANCEA2001 (E–W). (b)
Hypocenters of the same earthquakes projected onto the NW-SE vertical plane AB (dashed line in
a). DO, Dobrogea orogen; EEP, Eastern European platform; MP, Moesian platform; SP, Scythian
platform; and TB, Transylvanian basin. BOT, GRE, INC, LUC, OZU, PET, SCH, SEC, TES, and
VAR mark several sites where the probabilistic seismic hazard was assessed

about the features of earthquake ground motion excitation (source scaling), seismic
wave propagation (attenuation), and site effect in the region under consideration.
Ideally, all these factors should be studied on the basis of available regional
earthquake ground motion data.
The principal aims of this article are to analyze tectonic stress generation and
seismicity in the SE-Carpathians based on a vast body of new regional observations
and to present recent results of the regional seismic hazard analysis. Some of these
results, e.g., spectral models, site effects, and probabilistic seismic hazard assessment
(PSHA) for particular sites have been already described by Sokolov et al. (2004b, c,
2005). However, for a consistency of the analysis they are discussed here together
with new results on the site-dependent PSHA for eastern and southern Romania.

2 Geodynamic models and seismicity

The 1940 earthquake gave rise to the development of a number of geodynamic


models for this region. McKenzie (1972) suggested that this seismicity is associated
with a relic slab sinking in the mantle and now overlain by continental crust. The
1977 large earthquake and later the 1986 and 1990 earthquakes again raised ques-
tions about the nature of the earthquakes. A seismic gap at depths of 40–70 km
beneath Vrancea led to the assumption that the lithospheric slab had already
detached from the continental crust (Fuchs et al. 1979). Oncescu (1984) proposed
123
496 Nat Hazards (2007) 42:493–514

that the intermediate-depth events are generated in a zone that separates the sinking
slab from the neighboring immobile part of the lithosphere rather than in the sinking
slab itself. Linzer (1996) explained the nearly vertical position of the Vrancea slab as
the final rollback stage of a small fragment of oceanic lithosphere, and Girbacea and
Frisch (1998) assumed that the break-off, affecting only the crustal portion of the
slab, was followed by horizontal delamination of its lower portion. Most recently
Sperner et al. (2001) suggested a model of Miocene subduction of oceanic litho-
sphere beneath the Carpathian arc and subsequent soft continental collision, which
transported cold and dense lithospheric material into the mantle. While Linzer
(1996), Girbacea and Frisch (1998), and Sperner et al. (2001) consider the slab to be
oceanic, its origin is still under debate: whether the descending lithosphere is oceanic
or continental. Pana and Erdmer (1996) and Pana and Morris (1999) argue that there
is no geological evidence of Mesozoic oceanic crust in the Eastern Carpathians and
the descending lithosphere is likely to be thinned continental or transitional.
Continental convergence in the SE-Carpathians ceased about 10 Ma ago (Jiricek
1979; Csontos et al. 1992). Initially flat subduction zone began to steepen to its
present nearly vertical orientation (Sperner et al. 2005). At present the cold slab
(hence denser than the surrounding mantle) beneath the Vrancea region sinks due to
gravity. The hydrostatic buoyancy forces promote the sinking of the slab, but viscous
and frictional forces resist the descent. The combination of these forces produces
shear stresses at intermediate depths that are high enough to cause earthquakes.
This was shown in two-dimensional numerical experiments on tectonic stress evo-
lution by Ismail-Zadeh et al. (2000, 2005a). These authors recognized that the depth
distribution of the annual average seismic energy released in earthquakes has a
shape similar to that of the depth distribution of the predicted stress magnitude in
the slab. The models explain lateral compression in the slab as inferred from the
stress axes of earthquakes and the slab’s thinning and necking. The area of maximum
shear stress coincides with the region of high seismicity, and minimum shear stress in
the models is associated with the lower viscosity zone.

3 Regional seismic experiments

3.1 Teleseismic tomography

To reveal a relationship between the lithosphere dynamics and intermediate-depth


seismic pattern in the region, the international tomographic experiment CALIXTO
with 143 seismic stations was conducted in southeastern Romania in 1999 (Martin
et al. 2005). During the field experiment, 160 local events with magnitude Ml ‡ 2.0
and 450 teleseismic events with magnitude Mb ‡ 5.0 were recorded. The distance
between stations ranged from 15–20 km (the Vrancea region) to 25–30 km (outer
margins of the network), covering a region of about 350 km in diameter. The seismic-
tomographic model of the region consists of eight layers of different thickness (from
15 km to 50 km), which are each subdivided laterally into 42 · 42 km2 blocks
(Martin et al. 2005).
Data inversion reveals a high-velocity body with maximum P-wave velocity
perturbations of +3% in comparison with the background model (Fig. 2). This high-
velocity body is interpreted as the descending lithospheric slab. It reaches a depth of
at least 350 km (this is a maximum depth of high resolution tomography), which is in
123
Nat Hazards (2007) 42:493–514 497

Fig. 2 Seismic-tomographic image of the Vrancea slab (Martin et al. 2005) and hypocenters of
earthquakes (circles and asterisks indicate the location and magnitude of seismic events). The top
surface illustrates the topography. The blue surface represents the iso-surface of 3% positive
anomalies of P-wave velocity obtained via teleseismic data inversion. Focal spheres are fault-plane
solutions for the four largest Vrancea intermediate-depth earthquakes in the XXth century.
The right panel presents the horizontal slice of the seismic-tomographic image at depth 100 km

good agreement with results of previous seismic tomography studies (Oncescu 1984;
Wenzel et al. 1999; Bijwaard and Spakman 2000). The slab extends to the southwest
beneath the Moesian platform, however this portion of the slab is completely
aseismic (Fig. 2). Sperner et al. (2005) considered this aseismic portion to be already
delaminated from the overlying lithosphere.
Ismail-Zadeh (2003) found that the maximum shear stress migrates from the
upper surface of the slab to its lower surface in the course of changes in slab
dynamics from its active subduction through roll-back movements to sinking solely
due to gravity. The changes in stress distribution due to slab dynamics can explain
the present location of hypocenters of Vrancea events at the side of the slab (imaged
by the seismic tomography) adjacent to the Eastern European craton.

3.2 Seismic refraction studies

The crust and uppermost mantle above the Vrancea slab were imaged by two active-
source refraction seismic experiments carried out in 1999 and 2001 (Hauser et al.
2001, 2002). The 300 km long VRANCEA99 and the 460 km long VRANCEA2001
seismic refraction profiles crossed the Vrancea epicentral area in NNE–SSW and
ESE–WNW directions, respectively (Fig. 1). From forward and inverse ray trace
modeling, Hauser et al. (2001, 2002) distinguished a multi-layered crust with lateral
velocity variations in the sedimentary cover and minor changes in the crystalline
crust. They showed that the sedimentary succession comprises two to four seismic
layers of variable thickness with velocities ranging from 2.0 km s–1 to 5.8 km s–1. The
upper part of the seismic basement coincides with a velocity of 5.9 km s–1; velocities
in the upper crystalline crust are 5.9–6.2 km s–1. An intra-crustal discontinuity
apparent at depths between 18 km and 31 km divides the crust into an upper and a
lower layer.
123
498 Nat Hazards (2007) 42:493–514

The Moho discontinuity is predicted at a depth of about 40 km near the inter-


section of these profiles. Hauser et al. (2002) reported that the Moho shows no
crustal roots under the Carpathians. Velocities are 6.7–7.0 km s–1 within the lower
crust and about 7.9 km s–1 just below the Moho. Hauser et al. (2001) found a low-
velocity zone (7.6 km s–1) within the uppermost part of the mantle (at depths of
45–55 km) and the velocity beneath this zone is at least 8.5 km s–1. This low, velocity
zone is situated within the area of the seismic gap at depths of 40–70 km (Oncescu
1984).

4 Contemporary tectonic stress and crustal movements

To examine whether the areas of tectonic stress localization are consistent with the
areas of intermediate-depth seismicity in the region, we model the stress caused by
the sinking Vrancea slab in the SE-Carpathians. The model consists of (i) temper-
atures derived from a recent model of seismic P-wave velocity anomalies (Martin
et al. 2005) and surface heat flow, (ii) crustal and uppermost mantle densities con-
verted from P-wave velocities obtained from seismic refraction studies, (iii) geom-
etry of the Vrancea crust and slab from tomography and refraction seismic data, and
(iv) the estimated strain rate in the slab (as a result of earthquakes) to constrain the
model viscosity. We note that the tectonic stress model by Ismail-Zadeh et al.
(2005b) is based on the previous seismic-tomographic model by Martin et al. (2001),
where the variations in the crustal thickness was not considered in travel time cal-
culations. The seismic-tomographic model by Martin et al. (2005) is now used to
convert the seismic velocity anomalies beneath the region into temperature. We
follow the methodology by Ismail-Zadeh et al. (2005b) in modeling of synthetic
seismic velocities, considering the effects of anharmonicity (composition), anelas-
ticity, and partial melting on the seismic velocities. Once the synthetic velocities are
calculated for a first-guess temperature, an iteration process is used to find the ‘true’
temperature, minimizing the difference between the synthetic and ‘observed’ (in
seismic tomography experiments) velocities. The temperature in the shallow levels
of the region is constrained from measured surface heat flux corrected for paleo-
climate changes and for the effects of sedimentation (Demetrescu et al. 2001). We
use then the temperature so obtained to constrain the density and viscosity of
the model and compute tectonic stresses using a set of equations and boundary
conditions described by Ismail-Zadeh et al. (2005b).
The predicted maximum shear stress is associated with the high-velocity body,
localized at depths of about 70–170 km, and encompasses the area of major Vrancea
intermediate-depth events (Fig. 3). Horizontal compression at the intermediate
depths is consistent with the stress determination based on the focal mechanisms of
the earthquakes. Using numerous fault-plane solutions for intermediate-depth
events, Oncescu and Trifu (1987) showed that the axes of compressional stress are
almost horizontal (see Fig. 2).
An increase of shear stress due to the descending slab is one of the possibilities of
stress generation. Another process could be a plastic instability at high temperature,
when runaway shear slip (failure) occurs at even relatively low shear stresses (Griggs
and Baker 1969). Faulting due to metamorphic phase transitions (Green and
Burnley 1989) or dehydration-induced embrittlement (Raleigh and Paterson 1965;
Hacker et al. 2003) may also play a role in the regional stress generation and release.
123
Nat Hazards (2007) 42:493–514 499

Fig. 3 Maximum shear stress beneath the SE-Carpathians at different depths. Isolines present the
surface topography. Star marks the location of the Vrancea intermediate-depth earthquakes

Meanwhile, estimations of the cumulative annual seismic moment observed and


associated with the volume change due to the basalt-eclogite phase changes in the
Vrancea slab show that a pure phase-transition model cannot solely explain the
intermediate-depth earthquakes in the region (Ismail-Zadeh et al. 2000).
Geodetic studies on crustal motion have been performing since late 1990s. The
GPS network has been installed in the SE-Carpathians in 1997 and 1999. Mea-
surement campaigns in 1997, 1998, and 2000 revealed relative uplift rates in the
Vrancea region of about 10 mm yr–1 corresponding to an upper bound of 22 mm yr–1
(Dinter et al. 2001). Since no horizontal convergence is presently going on in the SE-
Carpathians, processes other than plate convergence are required to explain the
observed vertical movement in the region above the slab. The most plausible one is
isostatic rebound of the crust pulled down during subduction and collision (Ismail-
Zadeh et al. 2005a).

5 Quantitative seismic hazard assessment

5.1 Current state and available data

Analysis of the macroseismic and instrumental data from the intermediate-depth


Vrancea earthquakes reveal several peculiarities of earthquake effects (Mandrescu
1984; Mandrescu et al. 1988; Ivan et al. 1998; Mandrescu and Radulian 1999;
Moldovan et al. 2000). They may be summarized as follows: (i) the earthquakes
affect very large areas with a predominant NE–SW orientation; (ii) the local and
regional geological conditions can control the amplitudes of earthquake ground
123
500 Nat Hazards (2007) 42:493–514

motion to a larger degree than magnitude or distance; and (iii) the strong ground
motion parameters exhibit a large variability.
To assess the seismic hazard in the SE-Carpathians, strong ground motion exci-
tation and attenuation during the intermediate-depth Vrancea earthquakes were
analyzed (Marza and Pantea 1994; Lungu et al. 1995, 1999; Ivan et al. 1998; Oncescu
et al. 1999a; Moldovan et al. 2000; Radulian et al. 2000; Gusev et al. 2002). The
studies are based on the macroseismic data and the analog accelerograms of the
large 1977, 1986, and 1990 Vrancea earthquakes (Lungu et al. 1995; Oncescu et al.
1999a). The 1977 earthquake was recorded in Romania by only one accelerograph
located in Bucharest. The other strong events produced more than 30 free-field
records each.
Small and moderate size Vrancea earthquakes are recorded by the permanent (K2)
accelerometer network (Bonjer et al. 2000). The accumulated database consists of
hundreds of records obtained during many small and moderate earthquakes (M £ 6.0)
by several dozens of stations equipped by digital accelerographs. Many records were
also obtained in 1999 by a temporary network during the CALIXTO experiment
(http://www-sfb461.physik.uni-karlsruhe.de/pub/A2/Calixto/calixto99_en.html). The
accumulated database allows analyzing features of ground motion excitation and
propagation in the region, including site response analysis (e.g., Bonjer et al. 1999;
Wirth et al. 2003; Sokolov et al. 2004b, 2005).
Several studies were carried out to estimate the seismic hazard in Romania using
a probabilistic approach (Lungu et al. 1995, 1999; Musson 2000; Mantyniemi et al.
2003; Ardeleanu et al. 2005). The azimuth-dependent empirical attenuation models
evaluated from regional strong motion data were used in some of these studies;
however, variations of the local site response were not taken into account. Con-
sidering recent needs of earthquake engineering, which require local soil effects to
be included in PSHA, we employ for this study the recently developed site-dependent
technique (Sokolov et al. 2004c).

5.2 Research technique

The research technique is based on Cornell’s (1968) approach to the PSHA and
assumes that an earthquake occurrence is a stationary random process. The tech-
nique incorporates the effect of all potential earthquake sources and the seismic
activity rate assigned to them. The PSHA scheme (Fig. 4) is based on the Fourier
Amplitude Spectra (FAS) and allows for detailed consideration of site geology
(Sokolov 2000; Sokolov and Chernov 2001). PSHA results may be obtained in terms
of various ground motion parameters (PGA, RSA, and seismic intensity). However,
the approach and computational scheme differ somewhat from the Cornell’s (1968)
approach. The comprehensive description of the technique may be found in Chernov
(1989); Sokolov (2000); Sokolov and Chernov (2001); Jaiswal et al. (2004); Sokolov
et al. (2004c). Here we describe briefly the main steps of the approach.
Each potential earthquake is considered as an individual event, instead of the use
of the probability density functions for magnitude and distance in a seismic source
zone. At the first step of the computational procedure a log-normal distribution of
the ground motion parameter’s values is assumed for a given magnitude and source-
to-site distance. The probability, that the ground motion parameter does not exceed
a certain value, is then calculated for this single event (N = 1) with the given
magnitude M, depth H, and epicenter-to-site distance R:
123
Nat Hazards (2007) 42:493–514 501

Fig. 4 Schematic representation of the approach to seismic hazard assessment used in this study

Zx n o
1
PNðM¼m;R¼r;H¼hÞ¼1 ½X  x ¼ pffiffiffiffiffiffi exp ðx  aÞ2 =2r2x dx; ð1Þ
rx 2p
xmin

where X is the ground motion parameter; a is the mean value of log10X for
an earthquake of given M, R, and H; rx is the standard deviation; and xmin is a
sufficiently small value (xmin  a – 5rx).
If the FAS is evaluated as the ground motion parameter for a rock site condition,
the local site effect can be introduced by the frequency-dependent amplification
factor. To consider uncertainties of the site response, the spectral amplification
should be described as a random variable, and Eq. 1 may be rewritten as follows

X
Wmax

1
PNðM¼m;R¼r;H¼hÞ¼1 ½X  x ¼ pffiffiffiffiffiffi 
Wmin rx 2p
2 x 3 9 ð2Þ
Z =
4 expfðx  aÞ2 =2r2x gdx5P½W ¼ Wk  ;
;
xmin

where a = aR + log10 Y, aR is the FAS on a rock site, Y is the spectral amplification,


Ymin and Ymax are the minimum and maximum values of the spectral amplification,
respectively, and P[Y = Yk] is the probability that the spectral amplification equals
Yk. The introduction of Yk allows the use of different spectral amplification values for

123
502 Nat Hazards (2007) 42:493–514

small nearby and large distant earthquakes considering the peculiarities of the site
response that may depend on the intensity of the input motion and earthquake
characteristics (azimuth, earthquake depth, etc.) Most of the prevalent techniques
(when compared to the described technique) does not permit the consideration of site
effects before the hazard curve calculation, that is, the soil response is supposed to be
independent on the characteristics of the earthquake source and input motion.
At the next step of the computational procedure, we consider the distribution of
earthquake sources of the given magnitude and source-to-site distance through the
depth and earthquake occurrence. Finally, the influence of all potential earthquakes
of the given magnitude range located at various distances is taken into account.
We consider the aleatory variability, which is related to apparent randomness of
ground motion parameters during a single earthquake, introducing the probability
distribution function for the parameter at the first stage. Attenuation models for the
PGA and RSA at considered frequencies are evaluated using stochastic simulation
technique (Boore 1983, 2003). The PSHA in terms of seismic intensity (MSK scale)
is performed using the relationship between the intensity and FAS of the ground
acceleration proposed by Chernov (Chernov 1989; Chernov and Sokolov 1999) and
developed further by Sokolov (Sokolov 2002; Sokolov and Wald 2002). This rela-
tionship implies that the seismic intensity is determined by the level of ground
motion spectral amplitudes in the frequency range of 0.4–13 Hz. This approach
allows for reducing considerably the uncertainty in intensity—ground motion
parameter relations based on the peak amplitudes only (Chernov 1989).
We note that the conversion of the spectral amplitudes to other ground motion
parameters, including seismic intensity, is performed during the first stage of the
PSHA procedure based on mean amplitudes of FAS. Thus, all sources of the alea-
tory variability are considered, when calculating the probability that the ground
motion parameter does not exceed the certain value for a single event. Such a
conversion may introduce an additional uncertainty in predictions of the ground
motion parameters. However, our research technique allows to obtain site-dependent
assessment in terms of various ground motion parameters using a single model
described by Eq. 1.

5.3 Input data for probabilistic seismic hazard assessment

Evaluating seismic hazard of the Vrancea intermediate-depth earthquakes, we


consider all possible earthquakes that may occur within the Vrancea source zone. A
system of grid points (elementary cells) with 0.25 · 0.25 spacing is used. Earth-
quakes occur within a certain volume of the lithosphere, and the epicenters of
earthquakes are located at the center point of the cell. Each elementary cell is
characterized by the following parameters: (1) minimum (Mmin) and maximum
(Mmax) magnitudes of earthquakes which may occur in the cell; (2) probability
distribution of hypocentral depth for earthquake sources of Mmin £ M £ Mmax
(Mmin = 5 and Mmin = 8 are used in this study); and (3) rates of earthquake recur-
rence per unit time. We assume that all cells are characterized by the same Mmax and
by the same probability distribution of hypocentral depths (Fig. 5a).
We employ the ROMPLUS catalog (Oncescu et al. 1999b), which includes
moment magnitudes MW and is based on compiled earthquake data and on
corrected depth locations. The catalog is supposed to be complete for magnitudes

123
Nat Hazards (2007) 42:493–514 503

Fig. 5 Input data for PSHA (Sokolov et al. 2004c). (a) Depth distribution of hypocenters (1990–
2002). (b) Frequency-magnitude relationship for the Vrancea region for depths greater than 60 km.
(c) Examples of the VHR spectra calculated for earthquakes of different magnitudes occurred at
150 km depth for the case of the epicentral area (solid lines) and Bucharest city (dashed lines,
epicentral distance about 130 km); (d) Examples of amplification functions IVHR (mean values and 1
standard deviation limits) for two soil sites, K2-stations: PET (epicentral area) and BOT
(Bucharest). Number in parentheses shows the number of used records

MW ‡ 7 for the time interval between 1411 and 1800, for MW ‡ 6.5 from 1801 to
1900, for MW ‡ 5.5 from 1901 to 1935, for MW ‡ 4.5 from 1936 to 1977, and for
MW ‡ 3 from 1978 to 2003. However, the magnitude estimates before 1800 are
affected by large errors. Figure 5b shows the frequency-magnitude relation (the
average number of earthquakes of magnitude M ± 0.25 per 1 yr) for the depths
greater than 60 km. The relation used in this study is determined for the time
interval of 200 yrs for earthquakes of MW ‡ 6.25, and 100 yrs for 4.75 £ Mw £ 6.25.
The maximum moment magnitude for the Vrancea region was estimated to be 8
(Marza et al. 1991; Lungu et al. 1995; Mantyniemi et al. 2003). Earthquake recur-
rence parameters (the number of earthquakes of different magnitude per unit of
time and unit of volume) are assigned to the elementary cells as follows: the central
123
504 Nat Hazards (2007) 42:493–514

(axial) cells of the Vrancea region are characterized by the maximum number of
earthquakes, and gradually decreased numbers are assigned to the cells, which are
located toward the edge of the Vrancea region.
The specification of the input FAS model is of particular significance in our
approach. The general site-dependent model for the Fourier acceleration spectrum
A at frequency f is given by

Aðf Þ ¼ ð2p f Þ2 CSðf ÞDðR; f ÞIðf Þ; ð3Þ

where C is the scaling factor, S(f) is the source spectrum, D(R,f) is the attenuation
function, R is the source-site (hypocentral) distance, and I(f) represents the fre-
quency-dependent and intensity-(magnitude and distance)-dependent site response.
The FAS model determines a-value in Eq. 2 as a = A(f=f*) at frequency f *.
For the case of hypothetical ‘‘very hard rock’’ (VHR) spectrum, the function I(f)
should be equal to unity for the considered frequency range. Thus, the ratio between
observed spectra AO and the properly developed spectral model AVHR is considered
as the site-specific amplification function IVHR (Sokolov 1998; Sokolov et al. 2004a,
2005)

IVHR ðf Þ ¼ AO ðf Þ=AVHR ðf Þ: ð4Þ

The source function S(f) used in this study is represented as (Brune 1970, 1971)

Sðf Þ ¼ M0 =½1 þ ðf =f0 Þ2  ; ð5Þ

where the source acceleration spectrum at low frequencies increases as f2 and


approaches a value determined by f0 (corner frequency) and M0 (seismic moment) at
frequencies f > f0. The value of f0 is found from the relation f0 = 4.9 · 1010 b(Dr/
M0)1/3, where Dr is the stress parameter or stress drop, and b is the shear wave
velocity. Sokolov et al. (2004b, 2005) have recently shown that the simple x-square
spectral model (Brune 1970, 1971) can be used for VHR sites in Romania. In
general, the stress parameter Dr controlling high-frequency spectral amplitudes
increases with magnitude: 2–3 · 106 Pa for MW £ 3.5, 2–2.5 · 107 Pa for
4.8 £ MW £ 5.3, and up to 108 Pa for the case of large (6.0 £ MW £ 6.5) events.
The function D(R,f) accounts for the frequency-dependent attenuation that
modifies the spectral shape. It depends on the source-to-site distance, the regional
upper mantle and crustal material properties, and the frequency-dependent regional
quality factor Q that represents anelastic attenuation:
  
R
DðR; f Þ ¼ exp p f þj ; ð6Þ
Qðf Þb

where j is a site- and region-dependent parameter. The high-cut j-filter (Anderson


and Hough 1984) is introduced here to consider the near-surface (upper crust)
attenuation of seismic waves.
In the case of intermediate-depth Vrancea earthquakes, the generalized three-
layer Q-model is applied as Q(f) = Q0f 0.8, where Q0 = 150, 400, and 100 for the
depth ranges from 200 to 100 km, from 100 to 40 km, and from 40 to 0 km,
respectively (Radulian et al. 2000). Figure 5c shows examples of VHR spectra for
123
Nat Hazards (2007) 42:493–514 505

the case of the epicentral area and Bucharest city (epicentral distance about
130 km), which are calculated for the earthquakes occurred at depth 150 km.
The amplification function I(f) was estimated within a frequency range of 0.3–0.5
to 12–15 Hz using two ‘‘non-reference’’ techniques. One of the techniques is a
modification of the well-known horizontal-to-vertical Fourier spectral ratio (HVSR)
of the S-wave phase (Lermo and Chavez-Garcia 1993). In the second technique, the
site-specific amplification is evaluated using Eq. (4) (Sokolov 1998; Sokolov et al.
2004a, 2005). Basically the second technique is somewhat similar to the ‘‘empirical-
inferred’’ amplification technique (Atkinson and Cassidy 2000), in which the rock-
site motions are inferred from the analysis of all regional seismological data rather
than the data from a single reference station. We combine the techniques to analyze
a site-dependent seismic hazard, namely: the results of HVSR technique are
employed for low (f < 1.0 Hz) frequencies and VHR data are used for higher fre-
quencies. We do not consider possible non-linear response of soil during strong
excitation due to a lack of necessary information and data. Figure 5d shows exam-
ples of the amplification function. We note that these amplification function reflect
not only the soil conditions but also features of local geology up to basement.
A site effect is considered using two schemes. In the first scheme (Sokolov et al.
2004c), an analysis of the probabilistic site-dependent seismic hazard for Romania
was performed for particular points—locations of K2 accelerograph stations in
Romania (see Fig. 1). The local soil conditions for the stations vary from meta-
morphic rock to thick and water-saturated sedimentary formation. We consider
specific site amplification functions estimated for the given stations (it would be
impossible to do, if generalized attenuation relations are used).
The ground motion parameters are evaluated in the second scheme for areas in
eastern and southern Romania. The data from the K2-network and 36 sites of the
CALIXTO network are used to assess the generalized region-dependent site
amplification. The region under the study is divided into six characteristic segments
(Fig. 6a); the boundaries of the segments are chosen consistent with the geological
features and peculiarities of the distribution of the ground motion parameters.

Fig. 6 Evaluation of generalized site amplification. (a) Scheme of characteristic regions and the
location of K2 (circles) and CALIXTO (triangles) stations. Numbers in parentheses denote the
number of sites within the region. For the South-West region, the numbers show also the number of
stations (9) in the city of Bucharest. (b) Generalized region-dependent site amplifications (mean
amplitude values) including amplification for generalized ROCK category (Sokolov et al. 2005)
123
506 Nat Hazards (2007) 42:493–514

The generalized characteristics of site amplifications are evaluated by averaging the


amplification functions, which are obtained for all stations located within the given
segment. Mean amplitude values of the functions are shown in Fig. 6b. The average
amplification values may vary significantly from region to region depending on the
frequency range. The PSHA is performed for all nodes of the grid covering the
region, and each node is characterized by the relevant site amplification function.

6 Results of probabilistic seismic hazard assessment

Ground motion models, namely, values of site-dependent PGA and RSA for con-
sidered magnitudes and distances, are calculated based on the VHR spectral model
AVHR(f) and the frequency-dependent site amplification functions. Also we consider
uncertainties in the empirical amplification functions. PGA as well as PSA may be
sensitive to amplitudes of particular peaks in frequency-dependent soil amplification
functions. The procedure of averaging empirical amplification functions leads to
smoothing of the peaks. Therefore, the use of mean-amplitude amplification func-
tions in prediction of site-dependent PGA and PSA values may result in underes-
timation of the parameters.
Consideration of the uncertainty in site amplification in our approach is not a
trivial task for the case of peak amplitudes or response spectra. We use a simplified
procedure for this purpose. The values of PGA or PSA to be used as parameter X in
Eq. 1 are calculated using (i) a stochastic technique for the mean values of ampli-
fication and (ii) the mean +1 standard deviation amplitudes of empirical spectral
amplification. Technique (ii) is employed as a basis for the reasonable estimation of
the possible highest level of PGA or PSA.
Usually a value of the ground motion parameter with probability 0.1 of being
exceeded in 50 yrs (the average live period for conventional constructions) applies to
most building codes for ordinary buildings. It corresponds to 1/475 annual proba-
bility of being exceeded or to return period of 475 yrs. For buildings classified within
other categories of importance, different return periods may be applied (from 200 up
to 2,500 yrs). The new European standard (EUROCODE 8), besides return period
of 475 yrs, also recommends 95 yrs, i.e., 10% probability of being exceeded in 10 yrs.
Reference return periods from about 100–1,000 yrs are generally applicable to
electrical system components, and larger return periods (up to 10,000 yrs) are
applicable to dams, nuclear power plants, and other critical facilities. The models of
‘‘cumulative damage’’ require also assessments for short return periods. Meanwhile,
the short return periods (less than 50 yrs) may be useful to evaluate the hazard
assessment reliability by comparing with available ground motion recordings.
Therefore, in our calculations, we consider various return periods (from 50 to
2,000 yrs).
First, we consider the case when the PSHA has been performed for particular
sites (Sokolov et al. 2004c). Tables 1 and 2 present values of seismic intensity and
peak acceleration evaluated for various return periods or probability of being ex-
ceeded during specified exposure time. The amplitudes of ground motion are con-
trolled by the distance to the Vrancea seismogenic zone as well as by geological
factors, which can vary significantly due to variations of local soil conditions.
When Table 1 is inspected, the questions may arise: whether the seismic hazard
estimations are reliable and how to test the predictions? The best test is to compare
123
Nat Hazards (2007) 42:493–514 507

Table 1 PSHA results (seismic intensity, MSK scale) for the SE-Carpathians performed for
individual points (K2 stations) shown in Fig. 1

Station Epicentral Soil Return period T, yrs


distance
50 100 200 475 2000

VRI <40 Rock 7.2 7.8 (VIII) 8.0 8.5 9.7


GRE <40 7.5 8.4 (VIII) 8.7 9.4 10.2
PET 60 7.0 7.6 (VII–VIII) 7.8 8.2 9.7
SEC 60 7.5 8.3 (VIII–IX) 8.5 9.1 9.9
FUL 65 6.0 7.4 (VIII) 7.9 8.7 10.0
LUC 70 7.1 7.7 (VII) 8.0 8.5 9.7
OZU 80 Rock 6.4 7.3 (<VI) 7.7 8.0 9.5
SUL 90 5.6 6.8 (VII) 7.4 8.4 9.8
TUD 90 6.5 7.3 (VII–VIII) 7.5 8.0 9.3
AMR 110 5.3 6.4 (VI) 7.2 7.8 8.8
TES 110 Rock 4.6 6.2 (VII) 6.9 7.7 9.0
VAR 110 Rock 6.0 7.0 (VIII) 7.3 7.9 9.1
VOI 110 Rock 5.2 5.9 (<VI) 6.4 7.1 8.2
BDL 120 6.1 7.8 (VIII) 8.7 9.5 10.3
BOT 120 7.3 8.0 (VIII) 8.6 9.6 10.8
BVC 120 5.9 6.7 (VIII) 8.3 9.0 10.0
INC 120 6.0 7.2 (VIII) 7.7 8.5 9.9
BMG 130 6.6 7.4 (VIII) 7.8 8.7 9.8
CFR 130 Rock 6.0 6.7 (VII) 7.0 7.5 7.9
SCH 150 4.8 5.3 (VII) 5.6 6.0 7.0
TLC 180 Rock 5.0 5.6 (<VI) 6.2 7.0 7.8
MSA 180 5.4 7.1 (VI–VII) 7.5 7.8 8.1
HAD 180 Rock 6.4 7.3 (VII–VIII) 7.8 8.5 9.3
LTR 210 Rock 4.3 4.7 (<VI) 4.9 5.2 5.5

The values for T = 100 yrs are compared with the maximum intensity (in parentheses) observed
during last 120 yrs (Manderscu et al. 1988)

Table 2 PGA hazard for various return periods calculated for Vrancea and Bucharest

Area Return period, yrs PGA, cm s–2 Source of data

Vrancea 475 540 Mantyniemi et al. (2003)


390 Musson (1999)
300–600* Sokolov et al. (2004c)
100 320 Lungu et al. (2003)
150–300 Sokolov et al. (2004c)
Bucharest 475 290 Mantyniemi et al. (2003)
240–420* (360–600)a Sokolov et al. (2004c)
100 240 Lungu et al. (2003)
120–200 (180–310)a Sokolov et al. (2004c)

* Values may be overestimated, because the site response is considered to be linear during large
earthquakes
a
Mean + 1 standard deviation amplitudes of the site response are used (see text)

the actual effects from a large earthquake and the model predictions. As far as the
PSHA results are associated with many earthquakes, they determine the accepted
ground motion level expected during specified return periods. The PSHA results
should be, at least, consistent with the largest observed effect and should reflect the
123
508 Nat Hazards (2007) 42:493–514

general features of the ground motion distribution. Mandrescu et al. (1988) analyzed
large earthquakes, which occurred in Vrancea during 1893–1984, and developed a
map of the maximum observed intensity. Table 1 compares the Mandrescu et al.
(1988) results with our 100-yr return period estimation (40% probability of being
exceeded during 50-yr life time of constructions).
The results of PSHA for the SE-Carpathians illustrate a good agreement with the
observed effect. The lower (compared to the observed values) PSHA estimates are
obtained for two stations TES and VAR located at rock formation (Sokolov et al.
2004c), which reflects a difference between the seismic effect at soil and rock sites.
However, there are three cases conflicting with the observations: sites OZU and
LUC show much higher PSHA estimates and site SCH much lower estimate than the
observed values. The use of additional empirical data, e.g., ground motion data
collected near the sites or geological information, may reduce the discrepancy
between the observed and calculated intensity of earthquakes.
The maximum intensity for 100-yr return period (8.0–8.4 MSK) was estimated for
sites SEC, BOT (Bucharest), and GRE. The PSHA estimates for 200-yr return
period (22% probability of being exceeded during 50-yr exposure time) are 8.5, 8.6,
and 8.7 MSK, and for 475-yr return period (10% probability) are 9.1, 9.4, and 9.6
MSK, for these sites, respectively. The values are consistent with the maximum
observed intensity (about X MSK) during last 200–300 yrs (Marza 1996).
The present seismic code of Romania (P100-92, see Lungu et al. 2003) defines the
earthquake hazard by 50-yr mean recurrence interval event, which for the Bucharest
area implies the maximum PGA of 0.2 g (intensity of VIII MSK). The new code
(P100-2003) proposal, which is based on the results of probabilistic seismic hazard
assessment for the Vrancea source zone, defines the hazard for a 100-yr seismic
event. The PGA in the Bucharest area is estimated to be 0.24 g (Lungu et al. 2003).
The 100-yr PGA-hazard values for the territory vary from 0.12 to 0.2 g or from 0.18
to 0.31 g, when considering mean +1 standard deviation amplitudes of site response
(Sokolov et al. 2004c; see Table 2).
We use the PSHA results for individual sites in the SE-Carpathians as well as
the generalized site amplification functions (Fig. 6) to estimate the seismic hazard
for eastern and southern Romania. Figures 7 and 8 show results of the site-
dependent PSHA in terms of PGA and MSK intensity together with schemes of
PGA and MSK distribution during the large Vrancea earthquakes (MW = 7.4,
March 4, 1977; and MW = 7.2, August 30, 1986). There is a good agreement be-
tween observations and our site-dependent estimations both in the shape of con-
tours and absolute values.
For comparison, we show also the PGA hazard scheme calculated without con-
sideration of site response—for rock site conditions (Fig. 8). The PSHA results,
which do not consider geological factors, are inconsistent with the distribution of
ground motion amplitudes during earthquakes.

7 Conclusion

Stress generation and seismic hazard associated with large intermediate depths
earthquakes in the Vrancea region have been analyzed in this article. A knowledge
of modern geodynamics, seismicity, stress regime, and strong ground motion in the
Vrancea earthquake-prone region assists in seismic hazards estimations. These
123
Nat Hazards (2007) 42:493–514

Fig. 7 Comparison of MSK intensity distribution during two large Vrancea earthquakes (a: MW = 7.4, March 4, 1977 and b: MW = 7.2, August 30, 1986) and the
PSHA results evaluated for two return periods (c: T = 475 yr and d: T = 100 yr). The legend in the PSHA schemes shows maximum PSHA intensity

123
509
510

123
Fig. 8 Comparison of PGA (maximum of two horizontal components) distribution during the Vrancea earthquake (a: MW = 7.2, August 30, 1986) and the PSHA
results evaluated for two types of site conditions (b: rock and c, d: soil) and for two return periods (c: T = 100 yr and d: T = 475 yr). Numbers at the contours are
Nat Hazards (2007) 42:493–514

scaled in cm s–2
Nat Hazards (2007) 42:493–514 511

estimations, together with the assessment of the exposed values and their vulnerability
to the earthquakes, can improve seismic risk estimations for the SE-Carpathians.
Based on recent seismic tomography data and on the knowledge of geodynamic
evolution of the region, contemporary tectonic stresses beneath the SE-Carpathians
have been modeled. We have demonstrated a correlation between the location of
large intermediate-depth Vrancea earthquakes and the predicted localizations of
maximum shear stress in the descending lithosphere.
The PSHA based on the FAS has allowed estimating the site-dependent seismic
hazard for the SE-Carpathians caused by Vrancea earthquakes. Using the same
ground motion model (FAS), the procedure allows estimating various ground
motion parameters used in engineering practice and risk assessment, namely: seismic
intensity, PGA, and RSA. The spectral models as well as characteristics of site-
response on earthquake ground motion have been obtained based on the regional
data including several hundred records of small and large earthquakes. The PSHA
results are consistent with the general features of the observed earthquake effects in
the SE-Carpathians. Based on these results, we can conclude that geological factors
play an important part in the distribution of earthquake ground motion parameters
within the region analyzed. The results can be considered as a basis for compre-
hensive site-dependent PSHA analysis in the region and as a basis for a new seismic
code.

Acknowledgements The constructive comments and suggestions of two anonymous reviewers are
gratefully acknowledged. We are very thankful to M. Martin and F. Wenzel for useful discussions
on the regional seismic tomography and seismic hazard. This research was supported by the
Collaborative Research Center SFB 461 ‘‘Strong Earthquakes’’, French Ministry of Research, and
Russian Academy of Sciences.

References

Anderson J, Hough S (1984) A model for the shape of the Fourier amplitude spectrum of
acceleration at high frequencies. Bull Seism Soc Am 74:1969–1993
Ardeleanu L, Leydecker G, Bonjer K-P, Busche H, Kaiser D, Schmitt T (2005) Probabilistic seismic
hazard map for Romania as a basis for a new building code. Nat Hazard Earth Syst Sci 5:679–684
Atkinson GM, Cassidy JF (2000) Integrated use of seismograph and strong-motion data to deter-
mine soil amplification: response of the Fraser River delta to the Duvall and Georgia Strait
earthquakes. Bull Seism Soc Am 90:1028–1040
Bijwaard H, Spakman W (2000) Non-linear global P-wave tomography by iterated linearized
inversion. Geophys J Int 141:71–82
Bonjer K-P, Oncescu L, Rizescu M, Enescu D, Driad L, Radulian M, Ionescu M, Moldoveanu T
(2000) Source- and site-parameters of the April 28, 1999 intermediate-depth Vrancea earth-
quake: First results from the new K2-network in Romania, In: Book of Abstracts and Papers of
the XXVII General Assembly of the European Seismological Commission, Lisbon,
SSA-2-13-O, 53
Bonjer K-P, Oncescu M-C, Driad L, Rizescu M (1999) A note on empirical site responses in
Bucharest, Romania. In: Wenzel F et al (eds) Vrancea earthquakes: tectonics, hazard and risk
mitigation. Kluwer Academic Publishers, Dordrecht, pp 149–162
Boore DM (1983) Stochastic simulation of high frequency ground motion based on seismological
model of the radiated spectra. Bull Seism Soc Am 73:1865–1894
Boore DM (2003) Simulation of ground motion using the stochastic method. Pure Appl Geophys
160:635–676
Brune JN (1970) Tectonic stress and the spectra of seismic shear waves from earthquakes. J Geophys
Res 75:4997–5009

123
512 Nat Hazards (2007) 42:493–514

Brune JN (1971) Correction. J Geophys Res 76:5002


Chernov YK (1989) Strong ground motion and quantitative assessment of seismic hazard,
FAN Publishing House, Tashkent (in Russian)
Chernov YuK, Sokolov VYu (1999) Correlation of seismic intensity with Fourier acceleration
spectra. Phys Chem Earth Part A: Solid Earth Geodesy 24:522–528
Cornell CA (1968) Engineering seismic risk analysis. Bull Seism Soc Am 58:1583–1606
Csontos L, Nagymarosy A, Horvath F, Kovac M (1992) Tertiary evolution of the intra-Carpathian
area; a model. Tectonophysics 208:221–241
Demetrescu C, Nielsen SB, Ene M, Serban DZ, Polonic G, Andreescu M, Pop A, Balling N (2001)
Lithosphere thermal structure and evolution of the Transylvanian Depression—insight from new
geothermal measurements and modeling results. Phys Earth Planet Int 126:249–267
Dinter G, Nutto M, Schmitt G, Schmidt U, Ghitau D, Marcu C (2001) Three dimensional
deformation analysis with respect to plate kinematics in Romania. Rep Geodesy 2:29–42
Fuchs K, Bonjer K, Bock G, Cornea I, Radu C, Enescu D, Jianu D, Nourescu A, Merkler G,
Moldoveanu T, Tudorache G (1979) The Romanian earthquake of March 4, 1977. II.
Aftershocks and migration of seismic activity. Tectonophysics 53:225–247
Girbacea R, Frisch W (1998) Slab in the wrong place: Lower lithospheric mantle delamination in the
last stage of the Eastern Carpathian subduction retreat. Geology 26:611–614
Green HW II, Burnley PC (1989) A new self-organizing mechanism for deep-focus earthquakes.
Nature 341:733–737
Griggs DT, Baker DW (1969) The origin of deep-focus earthquakes. In: Mark H, Fernbach S (eds)
Properties of matter under unusual conditions. Wiley, New York, pp 23–42
Gusev A, Radulian M, Rizescu M, Panza GF (2002) Source scaling of intermediate-depth Vrancea
earthquakes. Geophys J Int 151:879–889
Hacker BR, Peacock SM, Abers GA, Holloway SD (2003) Subduction factory. 2. Are intermediate-
depth earthquakes in subducting slabs linked to metamorphic dehydration reactions? J Geophys
Res 108:2030, doi:10.1029/ 2001JB001129
Hauser F, Raileanu V, Fielitz W, Bala A, Prodehl C, Polonic G, Schulze A (2001)
VRANCEA99—the crustal structure beneath the southeastern Carpathians and the Moesian
Platform from a seismic refraction profile in Romania. Tectonophysics 340:233–256
Hauser F, Prodehl C, Landes M, the VRANCEA working group (2002) Seismic experiments target
earthquake-prone region in Romania. EOS, Trans Am Geophys Union 83(457):462–463
Ismail-Zadeh AT (2003) Modeling of stress and seismicity in the south-eastern Carpathians: Basis
for seismic risk estimation. In: Beer T, Ismail-Zadeh AT (eds) Risk science and sustainability.
Kluwer Academic Publishers, Dordrecht, pp 149–162
Ismail-Zadeh AT, Keilis-Borok VI, Soloviev AA (1999) Numerical modeling of earthquake flows in
the southeastern Carpathians (Vrancea): Effect of a sinking slab. Phys Earth Planet Inter
111:267–274
Ismail-Zadeh A, Mueller B, Wenzel F (2005a) Modeling of descending slab evolution beneath
the SE-Carpathians: Implications for seismicity. In: Wenzel F (ed) Perspectives in modern
seismology, LNES, Vol 105. Springer-Verlag, Heidelberg, pp 203–224
Ismail-Zadeh A, Mueller B, Schubert G (2005b) Three-dimensional modeling of present-day
tectonic stress beneath the earthquake-prone southeastern Carpathians based on integrated
analysis of seismic, heat flow, and gravity observations. Phys Earth Planet Inter 149:81–98
Ismail-Zadeh AT, Panza GF, Naimark BM (2000) Stress in the descending relic slab beneath the
Vrancea region, Romania. Pure Appl Geophys 157:111–130
Ivan IA, Enescu BD, Pantea A (1998) Input for seismic hazard assessment using Vrancea source
region. Rom J Phys 43:619–636
Jaiswal K, Sokolov V, Sinha R, Wenzel F, Chernov Yu (2004) Probabilistic seismic hazard assess-
ment for Mumbai (western India) area. In: Chen YT et al (eds) Earthquake hazard, risk,
and strong ground motion. Seismological Press, Beijing, pp 7–30
Jiricek R (1979) Tectonic development of the Carpathian arc in the Oligocene and Neogene. In: Mahel
M (ed) Tectonic profiles through the western carpathians. Geol. Inst., Dionyz Stur, pp 205–214
Lermo J, Chavez-Garcia FJ (1993) Site effect evaluation using spectral ratios with only one station.
Bull Seis Soc Am 83:1574–1594
Linzer H-G (1996) Kinematics of retreating subduction along the Carpathian arc, Romania. Geology
24:167–170
Lungu D, Demetriu S, Radu C, Coman O (1995) Uniform hazard response spectra for Vrancea
earthquakes in Romania. In: Proc. 10th European Conf. on Earthq. Eng., Balkema, Rotterdam,
pp 365–370

123
Nat Hazards (2007) 42:493–514 513

Lungu D, Cornea T, Nedelcu C (1999) Hazard assessment and site dependent response for Vrancea
earthquakes. In: Wenzel F et al (eds) Vrancea earthquakes: tectonics, hazard and risk mitigation.
Kluwer Academic Publishers, Dordrecht, pp 251–267
Lungu D, Aldea A, Demetriu S, Craofaleanu I (2003) Seismic strengthening and seismic instru-
mentation—two priorities for seismic risk reduction in Romania. In: Proc. of the First Inter-
national Conference on ‘‘Science and Technology for Safe Development of Lifeline Systems’’,
4–5 November 2003, Sofia, pp 1–24
Mandrescu N (1984) Geological hazard evaluation in Romania. Eng Geol 20:39–47
Mandrescu N, Anghel M, Smalbergher V (1988) The Vrancea intermediate-depth earthquakes and
the peculiarities of the seismic intensity distribution over the Romanian territory. St Cerc,
Geofiz, Geogr GEOFIZICA 26:51–57
Mandrescu N, Radulian M (1999) Macroseismic field of the Romanian intermediate-depth earth-
quakes. In: Wenzel F et al (eds) Vrancea earthquakes: tectonics, hazard and risk mitigation.
Kluwer Academic Publishers, Dordrecht, pp 163–174
Mantyniemi P, Marza VI, Kijko A, Retief P (2003) A new probabilistic seismic hazard analysis for
the Vrancea (Romania) seismogenic zone. Nat Haz 29:371–385
Martin M, Achauer U, Kissling E, Mocanu V, Musacchio G, Radulian M, Wenzel F and CALIXTO
Working Group (2001) First results from the tomographic experiment CALIXTO’99 in
Romania. Geophys Res Abst 3:SE1.02
Martin M, Ritter JRR, the CALIXTO working group (2005) High-resolution teleseismic body-wave
tomography beneath SE Romania—I. Implications for three-dimensional versus one-dimensional
crustal correction strategies with a new crustal velocity model. Geophys J Int 162:448–460
Marza VI (1996) Romania’s seismicity file: 1. Pre-instrumental data. Special Publicat Geol Soc
Greece 6:141–148
Marza VI, Pantea AI (1994) Probabilistic estimation of seismic intensity attenuation for Vrancea
(Romania) subcrustal sources. In: Proc. XXIV Gen. Assembly ESC, Athens, Greece, vol. III,
pp 1752–1761
Marza VI, Kijko A, Mantyniemi P (1991) Estimate of earthquake hazard in the Vrancea (Romania)
region. Pure Appl Geophys 136:143–154
McKenzie DP (1972) Active tectonics of the Mediterranean region. Geophys J R Astron Soc 30:109–185
Moldovan I-A, Enescu BD, Ionescu C (2000) Predicting peak ground horizontal acceleration for
Vrancea large earthquakes using attenuation relations for moderate shocks. Rom J Physics
45:785–800
Musson RMW (2000) Generalised seismic hazard maps for the Pannonian Basin using probabilistic
methods. Pure Appl Geophys 157:147–169
Oncescu MC (1984) Deep structure of the Vrancea region, Romania, inferred from simultaneous
inversion for hypocenters and 3-D velocity structure. Ann Geophys 2:23–28
Oncescu MC, Bonjer KP (1997) A note on the depth recurrence and strain release of large Vrancea
earthquakes. Tectonophysics 272:291–302
Oncescu MC, Trifu CI (1987) Depth variation of moment tensor principal axes in Vrancea
(Romania) seismic region. Ann Geophys 5:149–154
Oncescu MC, Bonjer KP, Rizescu M (1999a) Weak and strong ground motion of intermediate depth
earthquakes from the Vrancea region. In: Wenzel F et al (eds) Vrancea earthquakes: tectonics,
hazard and risk mitigation. Kluwer Academic Publishers, Dordrecht, pp 27–42
Oncescu M-C, Marza VI, Rizescu M, Popa M (1999b) The Romanian earthquake catalogue between
984–1997. In: Wenzel F et al (eds) Vrancea earthquakes: tectonics, hazard and risk mitigation.
Kluwer Academic Publishers, Dordrecht, pp 43–47
Pana D, Erdmer P (1996) Kinematics of retreating subduction along the Carpathian arc, Romania:
comment. Geology 24:862–863
Pana D, Morris GA (1999) Slab in the wrong place: Lower lithospheric mantle delamination in the
last stage of the Eastern Carpathian subduction retreat: comment. Geology 27:665–666
Radu C (1979) Catalogue of strong earthquakes occurred on the Romanian territory. Part I—before
1901; Part II—1901–1979 (in Romanian). In: Cornea I, Radu C (eds) Seismological studies on
the March 4, 1977 earthquake. Bucharest, Romania, pp 723–752
Radu C (1991) Strong earthquakes occurred on the Romanian territory in the period 1901–1990
(in Romanian). Vitralii 3:12–13
Radulian M, Vaccari F, Mandrescu N, Panza GF, Moldoveanu CL (2000) Seismic hazard of
Romania: deterministic approach. Pure Appl Geophys 157:221–247
Raleigh CB, Paterson MS (1965) Experimental deformation of serpentine and its tectonic
consequences. J Geophys Res 70:3965–3985

123
514 Nat Hazards (2007) 42:493–514

Sandi H (2001) Obstacles to earthquake risk reduction encountered in Romania. In: Lungu D, Saito
T (eds) Earthquake hazard and countermeasures for existing fragile buildings. Independent
Film, Bucharest, Romania, pp 261–266
Sokolov VYu (1998) Rough estimation of site response using earthquake ground motion records, In:
Proc. Second Int. Symp. on the Effects of Surface Geology on Seismic Motion (ESG 1998).
Yokohama, Japan, pp 517–522
Sokolov VYu (2000) Hazard-consistent ground motions: generation on the basis of Uniform Hazard
Fourier Spectra. Bull Seism Soc Am 90:1010–1027
Sokolov VYu (2002) Seismic intensity and Fourier acceleration spectra: revised relationship.
Earthquake Spectra 18:161–187
Sokolov VYu, Chernov YuK (2001) Probabilistic microzonation of the urban territories: a case of
Tashkent city. Pure Appl Geophys 158:2295–2312
Sokolov VYu, Wald DJ (2002) Instrumental Intensity Distribution for the Hector Mine, California,
and the Chi–Chi, Taiwan, Earthquakes: a Comparison of Two Methods. Bull Seism Soc Am
92:2145–2162
Sokolov VYu, Loh CH, Wen KL (2004a) Evaluation of generalized site response functions for
typical soil classes (B, C and D) in the Taiwan region. Earthquake Spectra 20:1279–1316
Sokolov VYu, Bonjer K-P, Rizescu M (2004b) Assessment of site effect in Romania during inter-
mediate depth Vrancea earthquakes using different techniques. In: Chen YT et al (eds)
Earthquake hazard, risk, and strong ground motion. Seismological Press, Beijing, pp 295–322
Sokolov VYu, Bonjer K-P, Wenzel F (2004c) Accounting for site effect in probabilistic assessment of
seismic hazard for Romania and Bucharest: A case of deep seismicity in Vrancea. Soil Dyn
Earthq Eng 24:929–947
Sokolov VYu, Bonjer K-P, Oncescu M, Rizescu M (2005) Hard rock spectral models for interme-
diate-depth Vrancea (Romania) earthquakes. Bull Seism Soc Am 95:1749–1765
Sperner B, the CRC 461 Team (2005) Monitoring of slab detachment in the Carpathians. In: Wenzel
F (ed) Perspectives in modern seismology, LNES, Vol. 105. Springer-Verlag, Heidelberg,
pp 187–202
Sperner B, Lorenz F, Bonjer K, Hettel S, Müller B, Wenzel F (2001) Slab break-off—abrupt cut or
gradual detachment? New insights from the Vrancea region (SE Carpathians, Romania). Terra
Nova 13:172–179
Wenzel F, Lorenz FP, Sperner B, Oncescu MC (1999) Seismotectonics of the Romanian Vrancea
area. In: Wenzel F et al (eds) Vrancea earthquakes: tectonics, hazard and risk mitigation. Kluwer
Publishers, Dordrecht, pp 15–25
Wirth W, Wenzel F, Sokolov V, Bonjer K-P (2003) A uniform approach to urban seismic effect
analysis. Soil Dyn Earthq Eng 23:735–758

123

View publication stats

You might also like