Earthquakes in Vrancea

1
Earthquakes in Vrancea

• M. Radulian

National Institute for Earth Physics
Bucharest mircea_at_infp.ro
2
OUTLINE

• Seismotectonics
• Seismicity patterns
• Source vs propagation effects
• focal mechanism
• seismic wave attenuation
• anisotropy
• Seismic cycle
• cycle characteristics
• numerical simulation
• Coupling between subcrustal and crustal seismic
  activities
• Conclusions

3

• Vrancea zone location
• Placed at the contact of three major tectonic
  units, Moesian Plate (MP), East-European Plate
  (EEP) and Tisza-Dacia Plate (TDP), Vrancea zone
  is an intra-continental seismic area.

4
Arc systems in Europe
5
Reconstruction of the Mediterranean evolution
(after Mantovani et al., 2002)
6
Seismotectonics Detachment model

• Vrancea - last remaining portion of the slab, in
  a process of finally detaching or having just
  detached.
• Progressive tearing of the slab from NW to SE as
  the oceanic plate is consumed and the continental
  margin enters the subduction zone.
• The age-progressive volcanism along the East
  Carpathians is presumed to be caused by the
  progressive tearing of the slab.
• The concentrated slab pull causes enhanced
  subsidence in the foreland (Focsani Basin).

7
Detachment process
Sperner et al., 2003
Girbacea Frisch, 1998
8
Delamination model
The oceanic lithosphere subduction ended some
time in the late Miocene, and since then a
portion of East European or Moesian platform
continental lithosphere has been delaminated
along a horizontal mid-lithospheric interface and
dripping down into the upper mantle. The
delaminated lithosphere migrated SE some 130 km
into its present position beneath Vrancea
(steepening the sinking lithosphere dip to near
vertical).
After Garbacea and Frisch (1998)
Inflow of asthenosphere into the gap between the
delaminated and unaffected lithosphere of the
subducting plate. It explains the magmatism and
extensional basins in the back-arc area.
9
Thermo-tectonic model (Cloetingh et al., 2004)

• Cross-section located in the bend zone of the SE
  Carpathians. Scenario in four successive steps,
  from stable to unstable subduction. The
  instability manifests in an acceleration of
  sinking to great depth due to gravity stretching,
  strength depth zonation and the formation of
  double subvertical high-velocity bodies down to
  200 km depths domain.

After Cloetingh et al. (2004)
10
Vrancea seismicity

• Romania is frequently hit by strong
  intermediate-depth earthquakes which occur
  beneath the SE Carpathian bending zone - the
  Vrancea area - between 60 km to 180 km depths in
  an unusual narrow epicentral region of 30 by 80
  km.
• During the last century, five major earthquakes
  occurred within the Vrancea zone
• October 6th 1908 (Mw 7.1),
• November 10th 1940 (MW 7.7),
• March 4th 1977 (MW 7.4),
• August 30th 1986 (MW 7.1),
• May 30th 1990 (MW 6.9).
• The most recent moderate shock occurred on
  October 27th 2004 (MW 5.8).
• Vrancea is the most concentrated seismic area in
  Europe. The moment release rate here is as high
  as the moment release rate of Southern California
  (Wenzel et al., 1998).

11
Romanian seismic network
12
Real time data exchange

• NIEP RETREIVES DATA FROM
• -GEOFON (GE_PSZ, GE_APE)
• -ITALY ( MN_AQU)
• - BULGARIA (MN_VTS)
• - IRIS (ANTO KIV)
• NIEP DELIVERS DATA TO
• - GEOFON
• - BULGARIA
• -R. CZECH (PRAGUE BRNO)
• ORFEUS
• DATA CENTER

13
Seismicity peculiarities in Vrancea

• The hypocentre distribution is close to a
  bidimensional one (along a NE-SW vertical plane)
• The background seismic activity is
  quasi-constant
• Inhomogeneous structure of the slab
• Predominant focal mechanisms with reverse
  faulting and a nearly vertical NE-SW fault plane,
  parallel with the seismicity distribution
• Periodicity of major shocks and the deficit of
  earthquakes at intermediate magnitudes, between
  asperity type (M 5) and large (M gt 7) events.

14
Seismic activity
15
Background seismicity (1974 2006)
Background activity (M gt 2.9) Time step 6
months
16
Seismicity

• Epicentre distribution of
• earthquakes produced between 1
• January 1995 and 31 May 2007,
• located by JHD algorithm.

17
Seismicity
Hypocentre distribution on two vertical cross
sections centred on Vrancea epicentral area
(reference point at 45.5oN, 26.5oE) for a
catalogue of earthquakes occurred between January
1976 and June 2007. The crosses are for
magnitudes 4-4.9, diamonds for magnitudes 5-5.9
and stars for magnitudes greater or equal 6.
18
Seismicity patterns The simple geometry of the
seismic activity in the subcrustal domain (h gt 60
km) shows a few remarkable features

1. the seismicity is sharply cut off at about 170 km
   depth (only one isolated event was reported
   below, at 208 km depth)
2. the seismic activity is dipping from NE toward SW
   in a vertical volume extremely narrow on the
   NW-SE direction
3. the density of earthquakes is by a factor of 5
   higher in the deeper part (around 140 km depth)
   than in the upper part (around 90 km depth) of
   the seismic active body.

19
The areas where smaller events (M below 5)
are preferentially generated are coincident with
areas where larger events (M above 5)
are generated. In other words, the probability
of generating a moderate-size earthquake is
increased for the regions where smaller events
where previously generated.
Our tests using different time windows show that
this feature is not dependent on time suggesting
the existence of a particular earthquake
generation process facilitating the faulting
process and stress release in these areas
(dehydration, fluid contribution, phase
transformation, or other). This specific process
is obvious in the lower part of the seismogenic
volume and is responsible for the inhomogeneous
and non random distribution of foci in Vrancea,
which cannot be explained by a simple mechanical
rupture process and associated stress transfer.
20
Seismicity Seismic activity distribution on
depth
After Martin et al. (2006)
21
Slab inhomogeneous structure

• Two active segments where the last 4 major
  Vrancea earthquakes were generated
• A in the upper part of the subducted lithosphere
  (around 90 km of depth)
• B in the lower part of the subducted lithosphere
  (around 140 km of depth)
• Key issue
• Can the zone separating the two segments generate
  large earthquakes or it behaves like a barrier
  (as it did for more than 100 years)?

22
Slab inhomogeneous structure

• Fragmentation at large scale is revealed by
  seismicity analyses and the distribution of large
  events afershocks.

23
Frequency-magnitude distribution

• The frequency magnitude distribution can in many
  cases be described using the equation
• log N a - bM,
• (Ishimuto and Ida, 1939 Gutenberg and Richter,
  1944),
• where N is the number of earthquakes having
  magnitudes larger than M,
• and a and b are constants. (b with depth, b with
  time, and b with
• magnitude).
• The frequency-magnitude relation was established
  for large earthquakes (M gt 6.0) in major regions
  of the world. Detailed seismicity studies have
  revealed that the b-value varies within the range
  (0.5, 1.5), that it is sensitive to the choice of
  the region boundaries, time interval, and
  magnitude range.

24
Frequency-magnitude distribution

• Variation of the b slope on depth in Vrancea
• Low b values indicate zones of high stress

25
Frequency-magnitude distribution

• Variation of the b
• slope on time
• 1900-2004

26
Fault-plane solutions
Mechanisms indicate tension in down dip direction
with variably oriented P axes. The structure is
clearly sinking into the mantle under the
tension. Note that the Vrancea zone is bounded
approximately to the NE and SW by the
Intramoesian and Trotus fault zones, major strike
slip terrain boundaries dividing the eastern and
western Moesian Platform and the East European
Platform.
Focal mechanisms of earthquakes (Mw gt 5.0,
19772001) from the CMT catalogue (Harvard, USA).
Note the elevated topography of the hinterland
(Transylvanian) and foreland basins of the
Eastern Carpathians, and Quaternary extensional
basins within the interior of the bend region.
Topographic data from USGS GTOPO 30 (Smith and
Sandwell, 1997). After Knapp et al.,
Tedctonophysics 2005.
27
Focal mechanism

• Compressive, thrust faulting regime. The fault
  plane solutions of the instrumentally recorded
  large earthquakes are remarkably similar. They
  typically strike NE-SW ( 220o) and dip 60o to
  70o to the NW. The slip angle is about 90o (i.e.
  down dip compression). The similarity of the
  fault plane solutions implies a particular
  radiation pattern and isoseismal curves strongly
  elongated on the NE-SW direction.
• predominance of reverse faulting with extension
  along the slab except a segment around 100 km
  depth (Oncescu and Trifu, 1987 Oncescu and
  Bonjer, 1997 Radulian et al., 2002 ).
• this kind of fault-plane solution is tentatively
  explained by a slab-pull down process which
  controls the kinematics of the system.

28
Focal mechanism

• The recent results obtained on the basis of the
  CALIXTO99 tomography experiment (Martin et al.,
  2006) outline a high-velocity body more extended
  than expected from the seismicity data alone.
  The high-velocity body goes down to about 350 km
  depth. Although the lower portion of the slab (h
  gt 180 km) seems to be not seismically active
  (only one isolated event was reported at 208 km
  depth see Figure 2), the density excess
  relative to the neighbouring asthenosphere
  material, is able to induce a strong pulling down
  force in the slab. This force is responsible for
  the reverse faulting with vertical extension
  which strongly influences the mechanism of
  Vrancea earthquakes, no matter of small, moderate
  or large size.
• To explain physically the presence of a force so
  strong to generate on average three major shocks
  (M gt 7) per century, we have to admit that the
  subducting lithosphere still present beneath
  Vrancea is attached to the overriding crust
  (Sperner et al., 2001), contrary to the
  hypothesis of complete detachment (Fuchs et al.,
  1979).

29
T principal axes projection
Mechanisms of recent earthquakes indicate
down-dip extension
Dip angles of T axes on the seismic plane change
gradually like a peacock
30
P principal axes projection
31
Macroseismic field 1802 event

• The largest event is considered the shock
  occurred in 26 October 1802 (MW 7.9), felt from
  Saint Petersburg and Moscow to Greek islands and
  Belgrade.
• Collapse of all the church towers in Bucharest
• a few churches entirely collapsed
• a lot of houses were destroyed
• the well-known construction in that time, the
  Coltea tower, was half destroyed.

32
Ground motion patterns Macroseismic field
10/11/1940, Mw 7.7
4/03/1977, Mw 7.4
33
1977 event
34
Ground motion patterns PGV distribution
Peak ground acceleration distribution for the
Vrancea earthquake of August 30, 1986 (M 7.1)
Peak ground acceleration distribution for the
Vrancea earthquake of October 27, 2004 (M 6.0)
35
Asymmetry in attenuation
36
Asymmetry Eastern Carpathians - Romanian Plain
37
Normalized Fourier spectra
38
Asymmetry Transylvanian Basin - Romanian Plain
39
Normalized Fourier spectra
40
Average spectra

• S wave (comp. N comp. E)
• Normalized spectra
• Averages for events 990620, 990629, 990713,
  990807, 991012

41
Peak ground S-wave velocity distribution
42
Source vs. path effectsThe strong attenuation of
the amplitude toward NW is not matching the
radiation of the typical Vrancea fault plane
solution.
43
Seismic Tomography of the Carpathian Arc
teleseismic data
For the entire depth domain of Vrancea
intermediate deep seismicity (60-200 km) the
tomography analysis shows a strong contrast in
structure across Carpathian Arc (foreland and
back-arc regions). Note, two different structures
are separated by a thin band oriented parallel
with the Vrancea seismicity.
Tomography image for 70-100 km depth domain
(after Martin et al., 2005)
44
Heat flow
Map of the heat flow map distribution (after
Demetrescu and Andreescu, 1994).
45
Geothermal profile across SE Carpathians arc
(after Andreescu and Demetrescu, 2001).
46
Anisotropy

• GPS measurements (Van der Hoeven et al., 2005)
  and seismic wave anisotropy. Anisotropy obtained
  from teleseisms by Ivan (2003) - orange lines and
  Achauer et al. (2001) - yellow and blue lines.
  The line length scales the delay time, while the
  orientation indicates the fast polarization
  direction. Broadband instruments belonging to the
  CALIXTO tomography experiment are considered.

47
Anisotropy

• The retrograde eastward motion of the slab and
  the rapid eastward motion of the Tisia-Dacia
  block over the past 16Ma., as indicated by the
  GPS measurements, let to specific sub-slab mantle
  flow pattern. The dominant stress pattern in the
  lower lithosphere is rather associated to slab
  retreat than slab break-off.
• The rays having back azimuths in the northwest
  mostly travel in the upper mantle wedge or within
  the slab itself resulting into relatively small
  delay-times and strike perpendicular anisotropy.
  Conversely, southeastward back azimuths are
  related to rays mostly traveling in the subslab
  mantle. Hence SKS anisotropy sub parallel to the
  strike of the slab refers to subslab mantle flow
  patterns.
• The polarization direction mimics remarkably the
  GPS displacement field, including the strong
  perturbation along the Trotus fault (compare
  directions at F09 station and F11 station in Fig.
  7). The match is broken in the Vrancea epicentral
  area where both strike-parallel and
  strike-perpendicular polarization directions are
  observed. This correlates with the complex upper
  mantle processes concentrated beneath this area.
• The results indicate a strong region of
  anisotropy in the South-East Carpathians (time
  delays of 1.5-2s), likely related to the
  subduction process.

48
Vrancea seismic cycle

• In the last 100 years, several seismic cycles can
  be identified in the Vrancea subcrustal domain.
  Each cycle is characterized by a shock larger
  than 6.5 and a relative lack of earthquakes at
  intermediate magnitudes (6 to 7). Possibly, the
  major shocks are generated by a percolation type
  process (Trifu and Radulian, 1991).
• We can consider four successive cycles since the
  beginning of the last century
• A 1900 1940
• B 1940 1977
• C 1977 1986
• D 1986 present

49
Vrancea seismic cycle
Successive cumulative processes in Vrancea as
revealed by Benioffs curves.
50
Crustal seismicity

• To obtain an accurate image of the seismic
  activity configuration in space, we selected all
  the events recorded since 1991 up to the present
  with at least 6 P-phase readings per event. We
  obtained finally a set of 323 selected shallow
  events (h lt 60 km) occurred on the Romanian
  territory. We used all the available digital
  stations, both from the permanent networks (the
  telemetered network of the National Institute for
  Earth Physics of Bucharest and the digital
  accelerometer network, installed and operated in
  cooperation with the University of Karlsruhe,
  Germany in the framework of CRC461 programme) and
  from the temporary network deployed during 1999
  within the seismic tomography experiment
  CALIXTO99 (Wenzel et al., 1998).
• We applied the JHD algorithm (Douglas, 1967) to
  locate the events. The algorithm is efficient in
  identifying possible systematic errors in data
  set (reading errors, inadequacies in station
  coordinates or clock, etc.) and provides station
  corrections. The location errors, measured as RMS
  of residuals, are in general below 0.5 s.

51
Coupling subcrustal/crustal seismicity

• Concentration of crustal seismicity at the
  Carpathians Arc bend, close to Vrancea, indicates
  a coupling between the processes below crust and
  in the crust.
• Epicentres distribution for selected shallow
  earthquakes (h lt 60 km) occurred between 1991 and
  2007 for different depth intervals empty
  circles, 010 km, solid circles, 1020 km, empty
  rectangles, 30-40 km, solid rectangles, gt 40 km.
  Hatched ellipsis roughly represents the
  epicentral area for Vrancea subcrustal
  earthquakes.

52
Coupling subcrustal/crustal seismicity

• A few remarkable seismicity patterns are
  noticeable in the figure
• (1) the cluster of earthquakes adjacent to the
  epicentral area of Vrancea intermediate-depth
  earthquakes (hatched ellipse) and partly
  overlapping this area
• (2) the seismicity along the major Intramoesian,
  Peceneaga-Camena, Trotus and Sf. Gheorghe faults
  
• (3) the lack of earthquakes between the
  Intramoesian and Peceneaga-Camena faults if we go
  away from Vrancea toward Black Sea and the lack
  of earthquakes in the back-arc region
  (essentially, no well located event is noticed in
  these areas).

53
Asymmetry/symmetry in seismicity configuration
54
Earthquake sequences in the foredeep

• The earthquake sequences which frequently occur
  at the contact between Focsani depression and
  orogen (so-called Râmnicu Sarat area) are
  following the same trend of thrust faults
  paralleling the orogenic front (Tarapoanca et
  al., 2003), as expression of coupling between the
  shortening and collision of orogen with East
  European/Scythian block and foreland (Ziegler et
  al., 1995).

Distribution of epicentres for the main
earthquake sequences generated after 1991 in the
Râmnicu Sarat area. The epicentres of the main
shocks are represented by red stars. In all cases
sequences develop on NE-SW direction.
55
Earthquake sequences in the foredeep

• Fault plane solutions of the main shocks of the
  earthquake sequence recorded in Râmnicu Sarat
  zone. Nodale planes assumed to be real rupture
  planes are marked by bold lines. P axis and T
  axis positions are shown by solid and empty
  triangles, respectively. The fault plane
  solution for the events of September 2005 are
  not well constrained.

In all cases, the aftershock distribution reveals
a preferential direction of migration, in good
agreement with the fault plane solution of the
associated main shocks. Thus, the aftershocks
line up approximately on the same direction with
the strike of the NW dipping nodal plane of the
fault plane solutions of the main shocks
56
Conclusions

• Earthquake distribution in Vrancea at
  intermediate depths is unusually confined in a
  narrow volume dipping almost vertically.
• Seismicity is located some 130 km SE from the
  Miocene suture position.
• Predominant reverse faulting with extension along
  the slab.
• Geometry of the fault plane solution for large
  shocks corresponds with the geometry of the
  seismicity pattern.

57
Particular geometry of the foci

• Trajectories of high density of foci where the
  local potential for earthquake generation is high
  which are persistent in time.
• Small and moderate events are correlated in space
  and time.
• Extreme concentration of the small and moderate
  earthquakes at intermediate depth could be
  explained by the presence of specific geophysical
  processes facilitating the earthquakes
  generation. For example, the presence of hydrous
  minerals earthquakes occur in subducting slabs
  where dehydration is expected, and they are
  absent from parts of slabs predicted to be
  anhydrous. Thus, the sharp cut off of seismicity
  below 160 km depth could be ascribed to the
  stopping of dehydration reaction at these depths.
  
• The tendency of earthquakes to occur at the slab
  margins.

58
Conclusions

• Earthquakes are more efficiently generated in the
  bottom part of the active volume.
• It looks like that the bottom part controls the
  entire process. If our hypothesis is correct, the
  strong earthquakes in the upper part of the
  sinking slab, such as 1977 earthquake or 1990
  earthquake are generated in response to the
  preceding large shocks below (1940 and 1986,
  respectively), and all the seismicity
  configuration at shallower depths (including the
  crust) should correspond to this particular
  trigger mechanism.
• Subsequently, we expect the future major event to
  be generated in the lower lithosphere (130 160
  km depth).

59
Conclusions

• The largest events are the most infrequent, but
  the most important to understand, since they
  control the evolution of the system and are the
  most destructive events. The detailed seismicity
  pattern analysis suggests that recognizable
  patterns of smaller, more frequent events can be
  used to detect the generation of the next major
  event although reliable and efficient forecasting
  of the largest events is still questionable.
• Numerical simulations to identify the
  characteristics of the preparation process of the
  strong subcrustal events originating in Vrancea
  region and analyze how they can be incorporated
  in a time-dependent seismic hazard assessment.

60
Slab pulling down effects

• The seismicity pattern in the crust shows a clear
  asymmetry across Vrancea intense to the SE and
  negligible to the NW. The intensification toward
  SE is controlled by the pulling down process of
  the lithospheric plate still attached to the
  crust together with the rolling back to the SE.
  The asymmetry is apparent in the map of the
  Bouguer anomaly distribution, with a positive
  anomaly due to the raise of the Moho
  discontinuity beneath the Transylvanian Basin and
  a minimum value corresponding to the subsidence
  beneath the Focsani Basin.
• As approaching the bending zone, the foci depth
  of the crustal earthquakes increases.
• The most distinctive property of the earthquake
  sequences in the Focsani Basin area is the
  orientation of the rupture direction parallel to
  the Carpathians Arc bend in Vrancea rupture
  tends to expand in a NE-SW direction. We assume
  that this property is a direct reflection of a
  specific and localized deformation process
  controlled by a slab-pull down process beneath
  Vrancea.
• The asymmetry in ground motion distribution
  around Vrancea seismic area may yield a strong
  test of slab detachment and continental
  lithosphere delamination hypotheses put forth to
  explain the unusual seismicity and volcanism of
  the Carpathian arc.

61
Conclusions

• Vrancea an atypical and unconventional case.
• Strong lateral inhomogeneities outlined by the
  tomography image, heat flow, seismic wave
  attenuation and thermal field.
• They suggest the eastward slab retreat and
  decoupling between underlying asthenosphere and
  the slab itself.