Geophys. J . Int. (1997)131, F9-Fl3
F A S T - T R A C K PAPER
Interplay between subduction and continental convergence: a threedimensional dynamic model for the Central Mediterranean
A. M. Negredo,' R. Sabadini' and C. Giunchi2
' Dipartimento di Scienze della Terra, Universita di Milano, via L. Cicognara 7,20129 Milano, Italy, E-mail: [email protected]
Dipartement des Sciences de la Matiere, Ecole Normale Superieure de Lyon, 46 All& d'Italie, 69364 Lyon Cedex 07,France
Accepted 1997 July 25.Received 1997 July 24;in original form 1997 June 10
SUMMARY
The Mediterranean region is attracting considerable attention due to the complexities
of its tectonic setting, which is considered, worldwide, a unique natural laboratory for
studying the occurrence of extensional tectonics in a general context of continental
convergence. The Tyrrhenian-Apennine system is controlled by the west-dipping
subduction of the Adria-Ionian lithosphere and by the near north-south convergence
between the African and Eurasian plates. We provide the first 3-D dynamic model of
the Central Mediterranean that quantifies the effects of subduction and convergence
on surface deformation, in simplified geometry. The axis of the model extends from
Sicily to the Alps along the subduction hinge line. A convergence rate of 1 cm yr-'
parallel to the subduction hinge has been applied to the Tyrrhenian block, in agreement
with global plate-motion models. Density contrasts within the slab cause the gravitational sinking and roll-back of the slab in the southern Tyrrhenian domain. Modelling
results show a gradual decrease of hinge retreat from south to north, with values
ranging between 8 and 2 mm yr-', indicating that the arcuate geometry of the hinge
line along the Italian peninsula is ultimately controlled by the interplay between
subduction and convergence. The pattern of vertical velocity along directions perpendicular to the hinge, with subsidence in the foredeeps and uplift at the eastern border
of the Tyrrhenian domain, is maintained along the whole Italian peninsula, with higher
values in the southern areas.
Key words: central Mediterranean, continental convergence, extension, model,
subduction.
INTRODUCTION
The goal of our work is to quantify the effects of convergence
between the African and Eurasian blocks and subduction in
the southern Tyrrhenian on surface deformation by means of
3-D dynamic models. In particular, we want to check if the
arcuate shape of the subduction hinge line could result from
a higher roll-back velocity of the slab under the southern
Tyrrhenian than under the northern sector of the Tyrrhenian
domain, and if vertical motions inferred from neotectonics
are consistent with the tectonic mechanisms under
investigation.
The rate of north-northwestwards motion of the African
Plate is about 1 cmyr-I (DeMets et al. 1990). Continental
convergence caused the formation of the Sicilian ApennineMaghrebian thrust belt and is responsible for a seismicity
0 1997 RAS
pattern with compressive axes parallel to the convergence
velocity, from western Sicily to Gibraltar (Anderson & Jackson
1987; Rebai, Philip & Taboada 1992; Pondrelli, Morelli &
Boschi 1995). The opening since Tortonian times of the
Tyrrhenian Sea (Fig. 1) in a compressional setting has been
explained by a mechanism of back-arc extension related to the
southward and eastward migration of the subduction zone
(Malinverno & Ryan 1986). 2-D subduction models have been
applied to the southern Tyrrhenian Sea (Giunchi et al. 1996),
while the combined effects of convergence and subduction
have been investigated by Bassi, Sabadini & Rebai (1997) by
means of 2-D models in the horizontal plane, where the effects
of subduction have been parametrized by suction forces applied
at plate boundaries. In the present analyses, the 3-D effects of
subduction and the tectonic forces acting at plate boundaries
are taken into account in a self-consistent fashion, although in
a simplified geometry.
F9
F10
A . M . Negredo, R. Sabadini and C. Giunchi
Crust
Asthenosphere
Lithosphere
Transition layer
Lower
mantle
Figure 1. The modelled area, in light grey, is superimposed on the tectonic map of the Central Mediterranean. The circles on the upper part of
the figure denote zero displacement of the lithosphere perpendicular to the boundary. The arrow at the southern boundary of the Tyrrhenian
domain denotes the velocity applied to this boundary to simulate the motion of the African plate. The modelled subduction hinge, in dark grey, is
simplified with respect to the real one, in black, which is characterized by an arcuate shape. The two thick dotted lines perpendicular to the
modelled subduction hinge indicate the two transects, in the north and in the south, along which the horizontal and vertical velocities are evaluated.
The geometry and boundary conditions of the 3-D model are portrayed in the lower pannel. V, denotes the convergence velocity applied to the
southern boundary, while the springs represent the restoring force imposed at the surface.
0 1997 RAS, G J I 131, F9-Fl3
MODEL DESCRIPTION
In this preliminary 3-D analysis, we have neglected the arcuate
geometry of the hinge line (black line in Fig. l), this geometry
being one of the outcomes expected from our study: we want
to understand, on quantitative grounds, if the combined effects
of subduction and convergence could deform the hinge line
into an arcuate shape consistent with the observed pattern. As
indicated in Fig. 1, we have rectified the hinge from the
southern border of the Calabrian Arc to the Alps, which makes
the modelled hinge line (grey line in Fig. 1) roughly parallel
to the direction of convergence of 17” west between the African
and Eurasian blocks derived from global plate-motion models
(DeMets et al. 1990). In this analysis we thus consider a
simplified geometry in order to emphasize the interplay
between subduction and convergence; complexities due to
more sophisticated geometries are left for detailed analyses
which are underway at present. Model predictions are compared with the general trends of observed subsidence and uplift
patterns in the foredeeps and in the Apenninic chain and with
the horizontal deformation pattern inferred from geological
studies.
To model the dynamic effects of convergence and subduction
we have used the finite-element code MARC, with a 3-D mesh
consisting of about 9000 elements. Fig. 1 illustrates the model
geometry and the boundary conditions. The depth of the
model is 1800 km, while the horizontal extension is 2700 km.
We have assumed a linear viscoelastic Maxwell rheology, with
viscosities of loz4Pa s for the crust, 5 x 10” P a s for the
lithospheric mantle, loz1Pa s for the asthenosphere and transition layer, and 3 x 10” Pa s for the lower mantle (Whittaker,
Bott & Waghorn 1992; Spada, Ricard & Sabadini 1992). The
elastic structure is based on the PREM reference model
(Dziewonski & Anderson 1981). We have chosen the Maxwell
viscoelastic rheology because it reproduces the long-term viscous properties of the lithosphere and, at the same time, its
plate-like behaviour well on short timescales. In the southern
part of the model we have assumed a 500 km deep slab with
a lateral extension of 250 km and a dip angle of 70°, in
agreement with tomographic images (Spakman 1990; Wortel
& Spakman 1992; Selvaggi & Chiarabba 1995). In this area
(indicated by white triangles in Fig. l ) , we have adopted the
same slab geometry and density structure as Giunchi et al.
(1996). The density anomalies within the slab, due to phase
transformations of a subducting oceanic plate, are based on
the petrological model of Irifune & Ringwood (1987), and
reach a maximum value of 4 0 0 k g m - 3 at 400km. In the
remaining part of the model, the interaction between the
Adriatic and the Tyrrhenian domain occurs via the
underthrusting of the Adriatic lithosphere, which reaches a
depth of 90 km, in agreement with the maximum seismicity
depth (Selvaggi & Amato 1992) (black triangles in Fig. 1,
lower panel). No density anomalies within the down-bending
Adriatic Plate have been considered in this area.
The bottom of the model is fixed in the vertical direction.
The buoyant restoring force is applied at the top of the model,
and is assumed to be proportional to the vertical displacement
and to the density contrast at the surface (Williams &
Richardson 1991). The density contrasts in the slab and the
convergence velocity are activated at time t = O and maintained
at a constant value thereafter, following the same scheme as
Whittaker et al. (1992). The unlocked subduction fault, based
0 1997 RAS, G J I 131, F9-Fl3
on the method of slippery nodes, allows for the gravitational
sinking and roll-back of the slab.
At the northern, western and eastern boundaries. the displacement of the lithosphere in the direction normal to the
boundary is made to vanish. At the northern boundary, this
condition accounts for the notion that convergence is considered with respect to a fixed northern Europe. The boundary
conditions in the west and east account for the finiteness of
the Mediterranean domain. In the southern boundary we have
applied a velocity of 1 cm yr-’ to the Tyrrhenian block i n the
direction parallel to the subduction hinge to simulate the
motion of the African Plate. Since it is not clear whether this
convergence velocity affects the Adria domain or whether the
motion of this microplate is independent of the motion of
Africa, convergence is applied solely at the southern boundary,
from the hinge line to the left side of the model, in agreement
with global plate-motion models and with the style of seismicity
from western Sicily to Gibraltar (DeMets ei al. 1990; Pondrelli
et a/. 1995). The velocity fields from the numerical models are
sampled along two transects perpendicular to the modelled
subduction hinge in the northern and southern sectors of the
Italian peninsula (thick dotted lines, Fig. 1, top panel). The
northern profile is located 400 km to the south of the Alps
and crosses the Northern Apennines, while the southern profile
crosses the modelled Calabrian arc.
The accuracy of the 3-D solution has been verified. prohibiting sideways flow and comparing the results with 2-D
models. After a time interval of about 250 kyr, dynamic
equilibrium between the restoring force and the forces associated with the density contrasts and convergence is attained
and the unrealistic initial stress distribution due to instantaneous loading has vanished and reached steady-state values;
horizontal and vertical velocities are then sampled at the
surface, providing us with valid estimates for 105-106 yr, during
which the geometric configuration does not change significantly. The comparison between the model-predicted velocity
at the surface and that inferred from geological indicators must
thus be taken with caution, not only because of the simplified
geometry and rheology, but also because of the timescale of
the geological observations, which is much longer than the
timescale of validity of our modelling.
MODELLING RESULTS
Fig. 2 shows the components of (a) the horizontal velocity
perpendicular to the subduction hinge and (b) the vertical
velocity along the two profiles portrayed in Fig. 1. Hereinafter,
the term ‘horizontal velocity’ must be understood as the
component perpendicular to the subduction hinge; the velocities are sampled at the surface of the model, with the horizontal
scale increasing from west to east.
The hinge line (at 1660 km) corresponds to a velocity
discontinuity due to the decoupling between the Tyrrhenian
and the Adria-Ionian domains. Positive values of horizontal
velocity denote an eastward motion. The hinge-retreat velocity,
which is calculated as the difference, at the hinge line, between
the horizontal velocities of the Tyrrhenian and Adria-Ionian
domains, is 8 mm yr-’ in the southern profile and 2 mm yr-I
in the northern profile (Fig. la). The decrease from south to
north, caused by the different depth and density structure of
the Ionian subduction with respect to the Adriatic Plate, is
consistent with geological indications (Patacca, Sartori &
A . M . Negredo, R. Sabadini and C. Giunchi
F12
8'""'"'"'"""'''~""'
Southern Profile
E 6.____
Northern Profile
E
v
h
L
(a)
-
P
.- - - ....-...--...
--....
-.._
0
.-N
8
r
.
-20
w
300
600
900 1200 1500 1800 2100 2400 2700
Distance (km)
E
Calabrian Arc (b)
Adriatic foredeep
0
w
300
600
900 1200 1500 1800 2100 2400 2700
Distance (km)
E
Figure 2. (a) Horizontal component of the velocity perpendicular to
the subduction hinge and (b) vertical component, in mm yr-' for the
southern (solid line) and northern (dashed line) profiles. The discontinuity at 1660 km corresponds to the hinge line, with the distance
measured from the left boundary of the model.
foredeep is induced by the sinking of the slab in the mantle,
while in the Adriatic foredeep it is due to overthrusting of the
Tyrrhenian block. The flexural response to this subsidence
causes the uplift of peripheral bulges at 1400 and 1900 km.
The pattern of vertical motions is the same for both profiles,
but with higher values for both subsidence and uplift in the
southern profile due to the slab pull, in agreement with the
neotectonic map of Italy (Ambrosetti et al. 1987). Along the
two profiles, the pattern of the vertical velocity is consistent
with the pattern of the topography at subduction zones
(Giunchi et al. 1996; Gurnis et al. 1996). Geological observations indicate a lower limit for the sedimentation rate during
the Pliocene of 1 mm yr-I in the Apulian foredeep (Royden,
Patacca & Scandone 1987) and a maximum value of
0.7 mm yr-' during the Quaternary in the Adriatic foredeep
(Kruse & Royden 1994). Modelling results show a decrease of
the subsidence rate from 1.5 mm yr-' in the southern profile
to 0.3 mm yr-' in the northern profile and are, therefore,
consistent with the observed variations from south to north.
The uplift rate at the outer arc of the overriding plate reaches
a maximum in the southern profile, in agreement with the high
uplift rate of 0.9-1.1 mm yr-' measured in the Calabrian arc
by Westaway (1993). Fig. 3 shows the variations of the horizontal and vertical velocities of the Tyrrhenian and AdriaIonian domains along the subduction hinge line. In order to
evaluate the effects of convergence we also show the results
obtained when only subduction is considered (V,= 0).
Convergence accounts for 50 per cent of the calculated hingeretreat velocity in the southern part of the model and causes
S
Scandone 1990; Cipollari & Cosentino 1994) and agrees well
with the actual shape of the hinge line, as shown in Fig. 1. Our
results clearly indicate that subduction is the necessary ingredient to produce the arcuate geometry of the hinge line
(Malinverno & Ryan 1986; Doglioni 1991). The deep and
heavy slab in the southern part of the modelled area enhances
the sinking of the Ionian lithosphere and the overthrusting of
the Tyrrhenian block. Furthermore, slab pull produces a
westward motion of the Ionian domain (southern profile),
whereas lateral extrusion due to convergence causes an eastward motion of the Adriatic domain (northern profile). The
low value of hinge-retreat velocity in the Northern Appenines
is consistent with the low rate of seismic strain release in this
area (Pondrelli et al. 1995). In the southern Tyrrhenian, we
obtain roll-back velocities lower than those inferred for standard continent-ocean interaction, where the domain is much
larger than ours and where the free-boundary conditions at
the edges of the plates allow for large relative velocities (Gurnis,
Eloy & Zhong 1996). The roll-back velocity predicted from
the model should be regarded as a lower bound due to the
plate-like behaviour of the viscoelastic lithosphere; a viscous
model, or more sophisticated non-linear rheologies (Spadini,
Cloetingh & Bertotti 1995), would speed up the velocity of
hinge retreat. Malinverno & Ryan (1986) estimated an average
velocity of hinge retreat in the Tyrrhenian of 2 cm yr-' for the
last 20 Myr, about a factor of two higher than ours. Bearing
in mind the intrinsic limitations of our modelling, such as the
simplified geometry and rheology, this difference suggests a
progressive reduction of roll-back velocity. However, this result
should be compared with ongoing GPS measurements. Vertical
velocities are shown in Fig. 2( b). Subsidence in the Apulian
N
-L 8
P
............ Tyrrhenian, V,
€ 6
E
v
-
.$
4
0
-
= 0 cmlyr
Adna-lonian, V, = 0 cm/
Adria-lonian V = 1 cm/
5
-2
ca
F o
.-8
z -2
I
0
200
400
600
800
1000
Distance (km)
S
1200
N
1
h
L
,x
E
E
- 0
.20
0
a,
>
- -1
.-
r
5
-2
0
200
400
600
800
1000
1200
Distance (km)
Figure 3. (a) Horizontal component of the velocity perpendicular to
the subduction hinge and (b) vertical component, in mm yr-' along
the hinge line showing a comparison of the results obtained when
both convergence and subduction are considered (V,= 1 cm yr-') and
when only subduction is modelled (Ii,=Ocmyr-'). The grey area
indicates the zone of deepest subduction.
0 1997 RAS, G J I 131, F9-Fl3
Subduction and convergence
lateral escape an d overthrusting of the Tyrrhenian block onto
the Adriatic domain, a n d outward extrusion of this domain.
Convergence is also shown t o have a noticeable effect on the
vertical velocity (Fig. 3b). Lateral extrusion of the Tyrrhenian
block o n t o the Adria-Ionian domain causes a n increase of the
uplift rate a t the border of the overriding plate and, as a
flexural response, extra subsidence in the Apulian a n d Adriatic
foredeeps.
CONCLUSIONS
O u r 3-D study sheds light on the effects of convergence and
subduction on the patterns of horizontal a n d vertical motion
in the Tyrrhenian an d Adria-Ionian domains. T h e most striking result is the decrease in the hinge-retreat velocity from
south to north, with values ranging between 8 a n d 2 m m yr-l.
This result supports the intuitive guess by Malinverno & Ryan
(1986) that the arcuate shape of the subduction hinge line
results from a faster roll-back velocity of the subducted Ionian
lithosphere t h an that of the northern areas, and agrees well
with recent results obtained from analogical models (Faccenna
e t al. 1996). O u r findings seem t o indicate a decrease in rollback velocity with respect to that in the geological past. Th e
pattern of vertical velocity along the direction perpendicular
to the subduction hinge line, with subsidence in the foredeeps
a nd uplift in the border of the overriding plate, is maintained
along the whole Italian peninsula, in agreement with the
neotectonic m a p of Italy (Ambrosetti e t al. 1987). T h e decrease
in the subsidence rate from the Apulian t o the Adriatic foredeep
a nd the maximum uplift rate in the Calabrian a r c are also in
agreement with neotectonics. Convergence accounts for 50 per
cent of the hinge migration velocity in the Calabrian Arc and
for the overthrusting of the Tyrrhenian block o n t o the Adriatic
plate in the central a n d northern sectors of the Italian
peninsula.
ACKNOWLEDGMENTS
This work was financially supported by the EU grant
‘Geodynamic modelling of the Western Mediterranean’
no. CHRX-CT94-0607. Rinus Wortel and an anonymous referee are thanked for their constructive remarks.
REFERENCES
Ambrosetti, P., Bosi, C., Carraro, F., Ciaranfi, N., Panizza, M.,
Papani, G., Vezzani, L. & Zanferrari, A., 1987. Neotectonic Map of
Italy, Progetto Finalizzato Geodinamica, Quad. Ricerca Scientifca,
4.
Anderson, H. & Jackson, J., 1987.The deep seismicity of the Tyrrhenian
Sea, Geophys. J. R . astr. Soc., 91, 613-637.
Bassi, G., Sabadini, R. & Rebai, S., 1997. Modern tectonic regime in
the Tyrrhenian area: observations and models, Geophys. J . Int.,
129, 330-346.
Cipollari, P. & Cosentino, D., 1994. Miocene tectono-sedimentary
events and geodynamic evolution of the Central Italy, in Neogene
Basin Evolution and Tectonics in the Mediterranean Area, pp. 28-29,
Abstr. Vol. Interim Colloquium Reg. Com. Medit. Neogene Sirat.,
Rabat.
0 1997 RAS, G J l 131, F9-FI3
F13
DeMets, C., Gordon, R.G., Argus, D.F. & Stein, S., 1990. Current
plate motions, Geophys. J . h i . , 101, 425-478.
Doglioni, C., 1991. A proposal for the kinematic modelling for westdipping subductions-possible application to the TyrrhenianApennine system, Terra Nonu, 423-434.
Dziewonski, A.M. & Anderson, D.L., 1981. Preliminary reference earth
model, Phys. Earth planet. Inter., 25, 297-356.
Faccenna, C., Davy, P., Brun, J.P., Funiciello, R., Giardini, D.,
Mattei, M. & Nalpas, T., 1996. The dynamics of back-arc extension:
an experimental approach to the opening of the Tyrrhenian Sea,
Geophys. J . Int., 126, 781-795.
Giunchi, C., Sabadini, R., Boschi, E. & Gasperini, P., 1996. Dynamic
modeis of subduction: geophysical and geological evidence in the
Tyrrhenian Sea, Geophys. J. Int., 126, 555-578.
Gurnis, M., Eloy, C. & Zhong, S., 1996. Free-surface formulation of
mantle convection-11. Implication for subduction-zone observables,
Geophys. J . Int., 127, 719-727.
Irifune, T. & Ringwood, A.E., 1987. Phase transformations in a
harzburgite composition to 26 Gpa: implication for the dynamical
behaviour of the subducting slab, Earth planet. Sci. Lett., 86,
365-376.
Kruse, S.E. & Royden, L.H., 1994. Bending and unbending of an
elastic lithosphere: the Cenozoic history of the Apennine and
Dinaride foredeep basins, Tectonics, 13, 278-302.
Malinverno, A. & Ryan, W.B.F., 1986. Extension in the Tyrrhenian
Sea and shortening in the Apennines as result of arc migration
driven by sinking of the lithosphere, Tectonics, 5, 227-245.
Patacca, E., Sartori, R. & Scandone, P., 1990. Tyrrhenian basin and
Apenninic arcs: kinematic relations since late Tortonian times, Mem.
SOC.Geol. It., 45, 425-451.
Pondrelli, S.A., Morelli, A. & Boschi, E., 1995. Seismic deformation in
the Mediterranean area estimated by moment tensor summation,
Geophys. 1. lni., 22, 938-952.
Rebai, S., Philip, H. & Taboada, A,, 1992. Modern tectonic stress field
in the Mediterranean region: evidence for variation in stress directions at different scales, Geophys. J. Int., 110, 106-140.
Royden, L., Patacca, E. & Scandone, P., 1987. Segmentation and
configuration of subducted lithosphere in Italy: an important control
on thrust-belt and foredeep-basin evolution, Geology, 15, 714-717.
Selvaggi, G. & Amato, A,, 1992. Subcrustal earthquakes in the
Northern Apennines (Italy): evidence for a still active subduction?
Geophys. Res. Lett., 19, 2127-2130.
Selvaggi, G. & Chiarabba, C., 1995. Seismicity and P-wave velocity
image of the Southern Tyrrhenian subduction zone, Geophys. J . lnt.,
121, 818-826.
Spada, G., Ricard, Y. & Sabadini, R., 1992. Excitation of true polar
wander by subduction, Nature, 360, 452-454.
Spadini, G., Cloetingh, S.A.P.L. & Bertotti, G., 1995.
Thermomechanical modeling of the Tyrrhenian sea: lithosphere
necking and kinematics of rifting, Tectonics, 14, 629-644.
Spakman, W., 1990. Tomographic images of the upper mantle below
central Europe and the Mediterranean, Terra Nova, 2, 542-553.
Westaway, R., 1993. Quaternary uplift of southern Italy, 1. geophys.
Res., 98, 21 741-21 772.
Whittaker, A,, Bott, M.H.P. & Waghorn, G.D., 1992. Stresses and
plate boundary forces associated with subduction plate margins,
J . geophys. Res., 97, 11 933-11 944.
Williams, C.A. & Richardson, R.M., 1991. A rheologically layered
three-dimensional model of the San Andreas fault in central and
southern California, J. geophys. Res., 96, 16 597-16 623.
Wortel, M.J.R. & Spakman, W., 1992. Structure and dynamics of
subducted lithosphere in the Mediterranean region, Proc. Kon. Ned.
Akad. v. Wetensch., 95, 325-247.