IC/96/93

United Nations Educational Scientific and Cultural Organizationand

International Atomic Energy Agency

INTERNATIONAL CENTRE FOR THEORETICAL PHYSICS

STRESS IN THE DESCENDING RELIC SLABBENEATH VRANCEA, ROMANIA

A.T. Ismail-ZadehInternational Centre for Theoretical Physics, Trieste, Italy

andInternational Institute of Earthquake Prediction Theory

and Mathematical Geophysics, RAS, Moscow, Russian Federation,

G.F. PanzaInternational Centre for Theoretical Physics, Trieste, Italy

andDipartimento di Scienze della Terra, Universita degli Studi di Trieste,

Trieste, Italy

and

B.M. NaimarkInternational Institute of Earthquake Prediction Theory

and Mathematical Geophysics, RAS, Moscow, Russian Federation,

MIRAMARE - TRIESTE

June 1996

ABSTRACT

We examine the effects of viscous flow, phase transition, and dehydration on the

stress field of the relic slab to explain the intermediate-depth seismic activity in the

Vrancea region. The finite-element model of the slab gravitationally sinking in the

asthenosphere predicts (1) the downward extension in the slab, as observed from

the stress axes of the earthquakes, and (2) the maximum magnitudes of stress to

occur in the depth range from 70 km to 160 km. The model is in good agreement

with the regional geological evolution. The depth distribution of the logarithm of

the annual seismic energy release has a shape only roughly similar to that of the

depth distribution of the stress magnitude in the slab, and the nonuniform time-

space distribution of the strong events in Vrancea cannot be explained solely by

viscous stress release. The study of the seismic moments of the intermediate-depth

earthquakes observed in the region indicates that a realistic mechanism lor triggering

these events in the Vrancea slab can be the dehydration of rocks, which makes fluid-

assisted faulting possible, rather than the shear stress caused by the basalt-eclogite

phase transformation in the oceanic slab.

INTRODUCTION

The Carpathians are bounded on the north and north-east by the Eastern Euro-

pean Platform, on the east and south by the Moesian Platform, on the west by the

Transylvanian and Pannonian basins. From the inspection of the earthquake epicen-

ters (Fig.l) and of the focal depths (Fig.2) in the area, a seismogenic body can be

revealed in the form of a parallelepiped with length of about 100 km, width of about

40 km, extending to a depth of about 160 km. Beyond this depth the seismicity ends

suddenly. This region, located at the Carpathians bend, is known as Vrancea.

As early as 1949, Gutenberg and Richter drew attention to the remarkable source

of shocks with foci in the depth range from 100 km to 150 km in Vrancea. According

to the historical catalogue [Novikova et al., 1995], the strongest intermediate-depth

shocks occur three or four times per century. In this century, strong events in the

depth range from 70 to 170 km occurred in 1940 with magnitude Ms=7.2 [Fuchs et

al., 1979], in 1977 Mw=7.5, in 1986 Mw=7.2, and in 1990 M^=6.9 (The Harvard

University Centroid-Moment Tensor Catalog).

McKenzie [1970, 1972] suggests that strong events in Vrancea occur in a vertical

relic slab subducting within the mantle and now overlain by the continental crust.

He believed that the origin of this slab is the rapid southeast motion of the plate

containing the Carpathians and the surrounding regions, relative to the Black Sea

plate. The Vrancea region is also considered [Fuchs et al., 1979] as a place where an

oceanic slab detached from the continental crust is sinking gravitationally.

Oncescu [1984] and Onccscu et al. [1984] propose a double subduction tectonic

model for Vrancea on the basis of the interpretation of a 3-D seismic tomographic

image. In their opinion, the decoupling of the sinking slab could be caused by the NW

push of the Black Sea plate. Using a large set of fault-plate solutions for intermediate-

depth shocks, Oncescu and Trifu [1987] show that the compressional axes are almost

horizontal and directed SE-NW, and that the tensional axes are nearly vertical, sug-

gesting that the slip is caused by the gravitational force.

According to these models, a cold, hence denser and more rigid than the surround-

ing mantle, relic slab beneath the Vrancea region sinks due to gravity. The hydrostatic

buoyancy forces help the slab to subduct, but viscous and frictional forces act as a re-

sistance to its descent. At intermediate depth these forces produce an internal stress

with one principal axis directed downward. Earthquakes occur in response to this

stress. These forces are not the only source of stress that leads to seismic activity in

Vrancea; in the process of slab descent the seismogenic stress can be caused by miner-

alogical phase changes and dehydration of rocks, which possibly leads to fluid-assisted

faulting.

The purposes of this paper are: (l) to study a simple numerical model of the

descending slab, in an attempt to explain the observed distribution of earthquakes;

(2) to examine the influence of the basalt-eclogite phase transition within the slab

on the stress in the surrounding rocks; and (3) to discuss a possible role of the

dehydration of rocks on the stress release within the descending slab in Vrancea.

VISCOUS STRESS IN THE DESCENDING SLAB

Introduction to the model

Numerical models of subducting slabs have been intensively studied by Vassiliou

et al. [1984] and Vassiliou and Hager [1988] to explain the global depth variation of

Benioff zones of seismicity. Here, to study the stress distribution and mantle flows

beneath the Vrancea region, we construct a model of the evolution of a relic oceanic

slab overlaid by the continental crust.

We assume that, keeping all the other parameters fixed, the number of earth-

quakes occurring in Vrancea at intermediate depths is related to the level of viscous

stress in the slab. We consider a simple model for the relic slab evolution and calculate

the stress therein assuming (1) that the Earth's mantle behaves as a viscous fluid at

the geological time scale, and (2) that the regional tectonic processes are associated

with mantle flows, regulated by Newtonian rheology.

The geometry and boundary conditions for the two-dimensional numerical model

used in the analysis are shown in Fig.3. A viscous incompressible fluid with variable

density and viscosity fills the model square (0 < x < L, ~H < z < k) divided into

four subdomains: atmosphere above z = 0, crust, subducting slab, and mantle. These

subdomains are bounded by material boundaries where density p and viscosity q a,re

discontinuous, while they are constant within each subdomain. The slab is modeled

as being denser than the surrounding mantle, and therefore tends to sink under its

own weight.

To test the stability of our results to variations of the density contrast, we consider

the value of 0.7 x 102 kg m~3, based on thermal models of the slab [Schubert et al.,

1975] and used in numerical modelling of subducting slab by Vassiliou et al. [1984],

and the value 0.4 x 102 kg m~3, suggested by the modelling of the long wavelength

component of Bouguer anomalies in correspondence of the lithospheric roots in the

Alps and in the Apennines [Werner and Kissling, 1985; Mueller and Panza, 1986;

Marson et al., 1995]. We also consider several values of the viscosity contrast between

the slab and the mantle: 0.4, 0.9, and 4.9 x 1021 Pa s, keeping the density contrast

equal to 0.4 X 102 kg m~3.

We solve Stokes' equation, which takes the following form in terms of the stream

function ij)

dp= py dxdxdz dxdz \dz2 dx2J \dz2 dx2 J dx

where g is the acceleration due to gravity. We assume impenetrability and free-flip

boundary conditions:

ij) = d24>jdx2 = 0 at x = 0 and x = L

if, = d2'<pjdz'i = 0 at z = -H and z = h.

The time-dependence of p and rj is described by the transfer equation

dA _ d$ dA di> dAdt dx dz dz dx

where A stands for p or 7]. The positions of the material boundaries, as functions of

time, are governed by the following set of two differential equations

dx/dt = d$ldz, dzjdt = -

where the points (x, z) are on the initial boundaries at £ = 0. The initial distributions

(t = 0) of p and r] and the positions of the material boundaries are known.

To solve the problem, that is, to compute the dependence of p, 7], material bound-

aries, velocity and stress on time, we employ an Eulerian finite element technique

described in detail by Ismail-Zadeh et ah [1994], Naimark and Ismail-Zadeh [1995],

and Naimark et al. [1996], We divide the model square into rectangular elements:

49 x 47 in the x and z directions. We use dimensionless variables, whereas in present-

ing the results for stress and velocity we scale them as follows: the time scale i*, the

velocity scale v*, and the stress scale a* are taken respectively as i* = r)*/[p*g(H + /*)],

v* = p*g(H+h)2/i]% and a* = p*g(H + h) where ii* = 1020 Pa s is a typical value of

mantle viscosity [Peltier, 1984] and p*=3.3 x 103 is a typical value of mantle density

[e.g., Turcotte and Schubert, 1982],

Numerical results

The parameter values used in the numerical modelling are listed in Table 1. We

choose 45° for the dip of the slab, at t = 0, as suggested by the distribution of

6

earthquake hypocenters in Vrancea (Fig.2); slight changes (±15°) in the initial slab's

dip yield results similar to those we describe here. The stress magnitude a is defined

as

= 10.5(^ + ^ + 2^ , ) ]^ = , 4J)xdzj \dz2 dx2 J

where T;J (i, j = x, z) are the components of the deviatoric stress.

The evolution of the slab that sinks under its own weight in the absence of external

forces is shown in Fig.4. for a density contrast 0.7 x 102 kg m~3 and a viscosity contrast

9 x 1O20 Pa s. The same computations made with a density contrast 0.4 x 102 kg m~3

show a nearly identical pattern. The numerical results show that variations of the

viscosity contrast lead to changes in the stress distribution and in the velocity of the

descending slab. If the viscosity contrast between the slab and the surrounding mantle

is as small as 0.4 x 1021 Pa s, then the stress in the slab is not large enough to give

rise to seismic activity in the region. A very high viscosity contrast (4.9 x 1021 Pa s)

causes a slow descent of the slab (about 0.3 cm yr"1) while the stress is now sufficiently

large to give rise to seismic activity. Our computations show that a viscosity contrast

of 9 x 1021 Pa s is more suitable for the Vrancea region, because in this case the

velocity of slab descent is about 1 cm yr"1, which agrees with the regional geological

observations [Bleahu et ah, 1973]. The subducting slab induces two mantle flows (Fig.

4 a-c). The flow on the left moves in clockwise direction, contributing to the evolution

of the Transylvanian basin and the folded arc. The other rotates counterclockwise,

and possibly affects the development of the Moldavian platform. Fig.4 d-f show the

axes of compression of the deviatoric stress. The axes of tension are perpendicular

to the axes of compression, and the magnitudes of tension and of compression are

the same. The maximum viscous stress is reached within the slab, and the axes of

compression are close to the horizontal direction, in agreement with the observed

focal mechanisms [Oncescu and Trifu, 1987].

The depth distribution of the average stress magnitude in the slab for two density

contrasts considered is represented in Fig.5. The two computed curves show that the

stress peaks in the depth range from about 70 km to 150 km. These curves have a

shape only roughly similar to that of the curve of log E versus depth, where E is

the annual seismic energy release (Fig.6). A close inspection of the curves in Fig.5

and Fig.6 shows that the maximum viscous stress is reached at a depth of about

90 km, whereas the maximum number of shocks is observed at a depth about 140

km. It is natural to assume that a higher level of stress causes an increased seismic

energy release per unit time, hence we have to consider other faulting processes at

intermediate depths.

INTERMEDIATE-DEPTH FAULTING PROCESSES

Strong earthquakes in Vrancea occur within a lithospheric slab sinking in the

asthenosphere. It is less obvious that the nonuniform time-space distribution of the

strong events in Vrancea might be explained solely by viscous stress release. High-

pressure faulting processes at intermediate depths in the Vrancea slab can also be

activated by the stress produced by heterogeneities in volume change, due to phase

transitions, and by the dehydration of rocks, which possibly leads to fluid-assisted

faulting.

Phase transition, seismic moment, and volume change

Slab metamorphism plays a crucial role in faulting processes at high pressures.

Many authors have considered the intermediate-depth earthquakes as the result of

phase changes from basalt to eclogite in the slab [e.g., Wiens et ah, 1993; Kirby

and Hacker, 1991; Cointe and Suarez, 1994]. There are two main effects of these

exothermic phase transitions (with a small positive Clapeyron slope): deflections of

the phase boundary from its normal position and release of latent heat. As for the

latter, it slightly changes the temperature of the surrounding material (Karato and

Sobolev, personal communication, 1995) and hence the buoyancy forces. Deflection of

the phase boundary depends upon the lateral temperature difference occurring in the

relatively cold slab that sinks into the hot mantle. The effects of the phase transition

in the slab have two implications for the stress state: (1) the denser phase acts as an

additional, load that pulls down the slab and causes an increase of the viscous stress;

(2) strongly exothermic polymorphic transformations in minerals under shear stress

tend to exhibit an unusual form of high-pressure failure [Kirby, 1995].

Experimental studies of Hacker et al. [1993] reveal that phase changes are very

sluggish under dry conditions, unless temperatures exceed 800-900° C are reached,

while wet rocks are readily transformed even at a temperature of about 600° C.

Hence Kirby [1995] infers that eclogite-forming reactions within a cold slab may be

delayed to the depth range from 70 km to 150 km, where dehydration occurs [Wilson,

1989]. As a region within a rock mass undergoes transformation to a denser phase,

contraction occurs in the direction of the maximum compressive stress, and within the

neighboring rocks la,rge deviatoric stresses are generated that lead to seismic failure.

To estimate the effect on the seismic activity in Vrancea, due to the volume change

associated with the basalt-eelogite phase transition, we employ the relation suggested

by McGarr [1977]

A ^n _ ™ pi - PQ

n=l fa

where MQ, is the seismic moment for the rath event caused by the volume change, p

is the shear modulus, / is the length of the slab along strike, T is the thickness of the

oceanic crust, vs is the velocity of descent of the slab, p0 is the density of rocks prior to

the phase transition, and p\ is the density of transformed rocks. Given fi = 6.5 x 1010

Pa [Turcotte and Schubert, 1982], / = 10s m, T = 104 m (the thickness of the typical

oceanic crust), vs = 10~2 in yr - 1 (the typical subduction rate), pa = 2.92 x 103 kg

m~3 (the typical density of wet basalts), pi = 3.5 x 103 kg m~3 (the typical density

of dry eclogites), we obtain the annual cumulative seismic moment of about 1017 N

m yr"1.

To estimate an observed seismic moment rate (OSMR) for events in Vrancea in the

depth range from 60 km to 170 km, we used the Harvard University Centroid-Moment

Tensor Catalog (a computer file, 1977-1995). This catalog contains events with M>5,

and sometimes with lower magnitude; the eight strongest shocks are listed in Table 2.

OSMR is found to be about 1.6 X 1019 N M yr"1 for the region. We consider a time

period of 19 years that includes almost all the strongest earthquakes that occurred in

the region for the last century. In the evaluation of OSMR, the time period considered

should be long enough to provide a representative sample of strong earthquakes in the

region. So, if a time interval is too short and does not include the strongest shocks,

it can result in an underestimate of OSMR and conversely one can overestimate the

moment rate, if the time window encloses an unusual sequence of strong events.

If we extend the time window to 1900, in the estimation of the annual OSMR,

we have to include the 1940 earthquake, with Ms=7.2 and hypocentral depth of more

than 100 km. It may be reasonable to assume for this event a moment equal to the

one of the 1977 event, and therefore, for this century, we get an OMSR of at least

5 X 1018 N m yr~1. this value can be representative of a longer period of time since

the strong earthquakes that occurred since 1600 seem to follow a regular pattern

[Novikovaet al., 1995].

Thus the cumulative annual seismic moment associated with the phase transition

is lower than that obtained from observations, and the pure phase-transition model

10

cannot explain the intermediate-depth seismicity in Vrancea.

Dehydration-induced faulting

Despite the fact that rocks in the subducting slabs are much more rigid than

the surrounding material, the frictional processes imposed by pressure prevent brittle

failure. At pressures above 3 GPa (about 100 km of depth) and even at a temperature

of 20° C brittle failure of rocks is impossible in absence of fluids [Green and Houston,

1995]. Raleigh and Paterson [1965] show on the basis of experimental investigations

that serpentinites (serpentinized peridotites) become brittle as a result of dehydration

at high pressures for which unhydrous rocks are plastically deformed.

It is well known from fracture mechanics that microcracks in a rock are generated

during brittle failure due to a tensile process. A fluid released by dehydration fills

the cracks and the pore fluid contributes together with the stress to the opening

of microcracks by filling them. As macroscopic stress continues to rise, the tensile

strength is exceeded, and finally, in some local region the rock becomes fractured so

that it loses its ability to support the compressive load, with the resulting formation

of a small fault within this region. The fault is bounded by a zone with a high density

of tensile microcracks. This zone filled by fluid thus becomes the principal seat of the

pore pressure generation that is necessary for fault growth.

Consequently, if a source of volatiles is available, there is a possibility for the

production of high-pressure faulting in the slab beneath Vrancea. Obviously, H2O is

carried down with the sediments covering the uppermost part of the slab, and the

hydrated oceanic crust contains about 2% of H2O at 3.0 GPa and 700° C. Moreover,

results of recent experimental studies [Ulmer and Trommsdorff, 1995] show that the

subduction of serpentinites containing about 13% of H2O may transport large quan-

tities of water to depths of the order of 150-200 km. Thus, the dehydration-induced

11

faulting in the depth range from 70 to 170 km can contribute to the increase of the

stress and consequently to the intermediate-depth seismicity observed in Vrancea.

DISCUSSION AND CONCLUSIONS

The Vrancea seismoactive region as well as the intracontinental regions of Burma,

Pamir-Hindu Kush, Central Apennines and Spain, are quite different from any other

subduction zone. Their distinguishing features are intermediate-depth events in pa-

leosubducted slabs. Studying the K2O/SiO2 ratio for the magmatic rocks. Boccaletti

et al. [1973] and Bleahu et al. [1973] suggest that the slab in Vrancea was subducted

during the Neogene time and reached depths of about 160 km where it partially

melted and generated calc-alkaline magmas, that erupted behind the Carpathian

folded arc, building up the magmatic arc. They also believe that the persisting sub-

duction caused an active stretching of the Transylvanian basin with the eruption of

the basaltic magma in the Quaternary. The finite-element model of a descending relic

slab allows us to explain the geological evolution and the seismic activity in Vrancea;

it is in agreement with the regional stress observed in the area: the axes of com-

pression and tension are close to the horizontal and vertical directions, respectively

[Oncescu and Trifu, 1987]. The model explains, even if roughly, the intermediate-

depth seismicity in the region if the seismic energy release depends exponentially on

stress.

The seismic moment rate due to the volume change, associated with the effect of

the basalt-eclogite phase transition in the descending slab, is much lower than that

obtained from the events which occurred in Vrancea in the depth range from 60 km

to 170 km. From this it follows that the basalt-eclogite phase transformation in the

descending slab is likely to have no fundamental effects on the production of stress

at intermediate depths in Vrancea. Alternatively, the generation of a pore fluid by

12

dehydration of hydrous minerals in the slab may give rise to dehydration-induced

faulting.

Thus, the viscous flows, due to the sinking of the relic slab together with the

dehydration-indxiced faulting can be considered as a plausible triggering mechanism

explaining the intermediate-depth seismicity in Vrancea.

Acknowledgments. The Harvard University Seismic Centroid-Moment Tensor

Catalog was kindly provided via Internet by the group of A. Dziewonski. We are

very grateful to V. Keilis-Borok, T. Kronrod, G. Molchan, R. Nicolich, A. Prozorov,

and I. Vorobieva for useful discussions of this research. We are also thankful to

R. Nicolich and L. Cernobori for the computing facilities at DINMA, University of

Trieste, and to I. Vorobieva for the preparation of some figures. The research was par-

tially carried out during the stay of A.T.I, at the University of Trieste. This work has

been supported by the U.S. National Science Foundation (grant # EAR 94-23818),

by NATO (Linkage grant # ENVIRLG 931206), and by the Russian Foundation for

Basic Research (grant # 95-05-14083).

13

REFERENCES

Bleahu, M.D., M. Boccaletti, P. Manetti, and S. Peltz, Neogene Carpathian Arc: A

continental arc displaying the features of an 'Island Arc', J. Geopkys. Res., 78,

5025-5032, 1973.

Boccaletti, M., P. Manetti, A. Peccerillo, and S. Peltz, Young volcanism in the

Calimani-Harghita mountains (east Carpathians): Evidence of a paleoseismic

zone, Tectonophysics, 19, 299-313, 1973.

Comte, D., and Suarez, G., An inverted double seismic zone in Chile: Evidence of

phase transformation in the subducted slab, Science, 263, 212-215, 1994

Constantinescu, L., and V. Marza, A computer-compiled and computer-oriented cat-

alogue of Romanian's earthquakes during a Millenium (984-1979), Rev. Roum,.

Geogr. Geophys., 193-234, 1980

Fuchs, K., K.-P. Bonjer, G. Bock, I. Cornea, C. Radu, D. Enescu, D. Jianu, A.

Nourescu, G. Merkler, T. Moldoveanu, and G. Tudorache, The Romanian earth-

quake of March 4, 1977. II. Aftershocks and migration of seismic activity,

Tectonophysics, 53, 225-247, 1979

Green, H.W. II, and H. Houston, The mechanics of deep earthquakes, Ann. Rev.

Earth Planet. ScL, 23, 169-213, 1995

Gutenberg, B., and C.F. Richter. Seismicity of the Earth and Associated Phenom-

ena. Princeton University Press, Princeton, 1949

Hacker, B.R., S.R. Bohlen, and S.H. Kirby, Albite —> jadeite + quartz transformation

in albitite, EOS, Trans. Am. Geophys. Un., 74, 611, 1993

14

Ismail-Zadeh, A.T., L.L Lobkovsky, and B.M. Naimark, Hydrodynamic model of

sedimentary basin formation based on development and subsequent phase trans-

formation of magmatic lens in the upper mantle, Computational Seismology and

Geodynamics, 26, 139-155, 1994

Kirby, S., Interslab earthquakes and phase changes in subducting lithosphere, Rev.

Geophys., Supplement, 287-297, 1995

Kirby, S., and B.R. Hacker, Intermediate-depth earthquakes, crustal phase changes

and the roots of arc volcauism, EOS, Trans. Am. Geophys. Un., 72, 481, 1991

Kondorskaya, N.V., Ed., Earthquakes in the USSR, 1962-1990. Nauka, Moscow,

1964-1992.

Marson, I., G.F. Panza, and P. Suhadolc, Crust and upper mantle models along the

active Tyrrhenian rim, Terra Nova, 7, 348-357, 1995.

Mueller, St., and G.F. Panza, Evidence of a deep-reaching lithospheric root under

the Alpine arc, in The Origin of Arcs, Developments in Geotectonics, 217 edited

by F.-C. Wezel, pp. 93-113, 1986.

McGarr, A., Seismic moment of earthquakes beneath island arcs, phase changes, and

subduction velocities, / . Geophys. Res., 82, 256-264, 1977.

McKenzie, D.P., Plate tectonics of the Mediterranean region, Nature, 226, 239-243,

1970

McKenzie, D.P., Active tectonics of the Mediterranean region, Geophys. J. Astron.

Soc, 30, 109-185, 1972

Moldoveanu, C.L., Novikova, O.V., Vorobieva, LA. and Popa, M., The updated

Vrancea seismoactive region catalogue, ICTP Internal report IC/95/104, 1995

15

Naimark, B.M., and A,T. Ismail-Zadeh, Numerical models of subsidence mechanism

in intracratonic basins: Application to North American basins, Geophys. J. Int.,

123, 149-160, 1995

Naimark, B.M., A.T. Ismail-Zadeh, and W. Jacoby, Numerical approach to problems

of gravitational instability of geostructures with advected material boundaries,

submitted to Geophys, J. Int., 1996

Novikova, O.V., Vorobieva, LA., Enescu, D., Radulian, M., Kuznetzov, I. and Panza,

G.F., Prediction of the strong earthquakes in Vrancea, Romania, using the CN

algorithm, PAGEOPH, 1^5, 277-296, 1995.

Oncescu, M.C., Deep structure of the Vrancea region, Romania, inferred from simul-

taneous inversion for hypocenters and 3-D velocity structure, Ann. Geophysicae,

2, 23-28, 1984

Oncescu, M.C., V. Burlacu, M. Anghel, and V. Smalbergher, Three-dimensional

P-wave velocity image under the Carpathian arc, Tectonophysics, 106, 305-319,

1984

Oncescu, M.C., and C.I. Trifu, Depth variation of moment tensor principal axes in

Vrancea (Romania) seismic region, Ann. Geophysicae, 5, 149-154, 1987

Peltier, W.R., The rheology of the planetary interior, Rheology, 28, 665-697, 1984

Raleigh, C.B., and M.S. Paterson, Experimental deformation of serpentine and its

tectonic consequences, J. Geophys. Res., 70, 3965-3985, 1965

Schubert, G.,D.A. Yuen, and D.L. Turcotte, Role of phase transitions in a dynamic

mantle, Geophys. J. R. Aslron. Soc, 42, 705-735, 1975

16

Trifu, C.L, and M. Radulian, A depth-magnitude catalog of Vrancea intermediate-

depth microearthquakes, Rev. Roum. Geophys., 35, 35-45, 1991

Turcotte, D.L., and G. Schubert. Geodynamics: Application of Continuum Physics

to Geological Problems. John Wiley & Sons, New York, 1982

Ulmer, P., and V. Trommsdorff, Serpentine stability to mantle depths and

subduction-related magmatism, Science, 268, 858-861, 1995

Vassiliou, M.S., B.II. Hager, and A. Raefsky, The distribution of earthquakes with

depth and stress in subducting slabs, / . of Geodynamics, 1, 11-28, 1984

Vassiliou, M.S., and B.H. Hager, Subduction zone earthquakes and stress in slabs,

PAGEOPH, i^S, 547-624, 1988

Werner, D., and E. Kissling, Gravity anomalies and dynamics of the Swiss Alps,

Tectonophysics, 111, 97-108, 1985

Wiens, D.A., J.J. McGuire, and P.J. Shore, Evidence for transformational faulting

from a deep double seismic zone in Tonga, Nature, 36^, 790-793, 1993

Wilson, M. Igneous Petrogenesis. Chapman & Hall, London, 1989. 452 p.

17

TABLE 1. Nomenclature

Notation Meaning Value

g acceleration due to gravity, m s~2 9.8

h height over the surface, km 33

hc initial thickness of the crust, km 40

hs initial thickness of the slab, km 30

H -\-h vertical size of the model, km 333

I length of the slab along strike, m 10s

L horizontal size of the model, km 350

MQ seismic moment for the nth event, N m yr"1

t time

t* time scale, yr 300

T thickness of the oceanic crust, in 104

v* velocity scale, in yr"1 1.1 x 103

vs velocity of slab subdnction, m yr"1 10~2

x horizontal coordinate of the model grid system

z vertical coordinate of the model grid system

rf typical value of viscosity, Pa s 1O20

7]aiT viscosity over the surface. Pa s 101D

fjc viscosity of the crust, Pa s 1023

r/m viscosity of the mantle , P a s 1O20

r]s viscosity of the slab, Pa s 1021

\i shear modulus, N m~2 6.5 X 1010

18

p* typical value of density, kg m~3 3.3 x 103

paiT density over the surface, kg m~3 0

pc density of the crust, kg m~3 2.9 X 103

p m density of the mantle, kg m~3 3.3 X 103

ps density of the slab, kg m~3 3.37 X 103 and 3.34 x 103

p0 density of rocks prior to the phase change, kg m~3 2.92 x 103

pi density of transformed rocks, kg m~3 3.5 X 103

a stress magnitude

CT* stress scale, Pa 1.1 x 101D

T{j deviatoric stress components

if) stream function

TABLE 2. Subcatalog of the strongest intermediate-depth earthquakes in Vrancea

beginning with 1977 event

No

1

2

3

4

5

6

7

8

Date

m/ d/ y

3/04/77

10/02/78

5/31/79

9/11/79

8/01/85

8/30/86

5/30/90

5/31/90

Time

h: m: s

19:21:54

20:28:53

07:20:06

15:36:54

14:35:03

21:28:36

10:40:06

00:17:48

Latitude

°N

45.77

45.72

45.54

45.56

45.74

45.54

45.86

45.79

Longitude

°E

26.76

26.47

26.32

26.29

26.50

26.29

26.67

26.75

Depth

km

84

154

114

143

103

133

74

87

Mo

N m yr"1

1.99 x 1O20

4.75 x 1016

7.26 x 1016

6.23 x 1016

7.96 x 1016

7.91 x 1019

3.01 x 1019

3.23 x 1018

19

FIGURES

Fig. 1. Epicenters of 1461 Romanian earthquakes from January 3, 1900 to March 31,

1995. Several catalogs have been combined to prepare Figs.l, 2, and 6 [Constantinescu and

Marza, 1980; Kondorskaya, 1964-1992; Trifu and Radulian, 1.991; Moldoveanu et al., 1995;

USGS-NEIS catalog; Harward University CMT catalog]

Fig. 2. Hypocenters of the same Romanian earthquakes (epicenters presented in Fig.l)

projected on the vertical plane along the W-E direction. Heavy squares stand for groups of

hypocenters located close together.

Fig. 3. Geometry of the model with the boundary conditions used in the calculations.

The z axis is directed upward (z = 0 corresponds to the earth's surface) and the x axis is

directed from left to right

Fig. 4. Flow fields (a-c) and deviatoric compression axes (d-f) for the evolution of

the slab subject only to gravitational forces: (a,d) t=16 Ma BP, (b,e) t=10 Ma BP, (c,f)

present-day. The maximum values of velocity and stress magnitude are represented at the

top of the figures

Fig. 5. Depth distribution of the average stress magnitude in the slab for density

contrasts 0.4 x 102 kg m~3 (1) and 0.7 x 102 kg m~3 (2)

Fig, 6. Logarithmic depth distribution of the annual seismic energy release E (mea-

sured in lCTr J) in Vrancea from 1962 to 1996. Events are grouped in 5 km intervals

20

45N 45N

44N25E 26E Z7E 28E

44N

MAGNITUDESUmax

0 km <= DEPTH60 km <= DEPTH

0.0 -•

< 60< 999

4.9

kmkm

5.0 - 5.9

•190D.

6.0 -D

1. 1-1999

•

6.9

.12.31

7.0

a7.9

Fig. 121

w

Q<DQ

2000

• I • I C I « B •% D

• • •* • • D• • • D i

DC -TI. , D •

Length, km 160

. M<4; n 4<M<5; D 5<M<6; D 6<M<7; D M>7

Fig. 222

oII

-hi0

no

v2=0xz

=0

Fig. 323

CO

t

1o

. i i i .

o

-10 o

-15

i I | I I

o

-20

i i I i io

-25

i < |o

-30

(Hld3Q

Fig. 4

24

o

I 1 1 1 1 1 1 1 1 10 0 0o in oT - T - CM

'Hid3Q

us

0in

I I 1 1 \ ~ o0oCO

Fig. 425

o

oo

LJJ

1 1ooT

1 1 1 •om

ooCM

oIS)CM

Fig. 426

COD_

o

LU

CDt

X

oLU

! i 'oi n

1^1 1 i ]oo

i l l io

1 1 I •oo

1 1om

Fig. 427

0

oLLJ

s

o1

ooT—

oID1 —

ooO o

o

Fig. 428

COQ_00oLLI

CD

LU

oin oo

oLO ooC\J

o<N

ooCO

Fig. 4

29

(IICl-

wCD

•4—•

CO

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

-

- /\j

- /Jy-rS-y,-''

-

/ V 2

I* \

; ' 1

i

V

\

I

-

-

-

\ \ —*

1 :-200 -150 -100

Depth, km-50 0

Fig.5

30

LJJD3O

-150 -100Depth, km

-50 0

Fig. 6

31