Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Geophysical Journal International

Geophys. J. Int. (2016) 205, 864–875 doi: 10.1093/gji/ggv533


GJI Geodynamics and tectonics

Lateral detachment in progress within the Vrancea slab (Romania):


inferences from intermediate-depth seismicity patterns

Horia Mitrofan,1 Mirela-Adriana Anghelache,1 Florina Chitea,1,2 Alexandru Damian,2


Nicoleta Cadicheanu1 and Mădălina Vişan1
1 Institute of Geodynamics, Romanian Academy, Jean-Luis Calderon 19–21, 020032 Bucharest, Romania
2 Department of Geophysics, Faculty of Geology and Geophysics, University of Bucharest, 020956 Bucharest, Romania. E-mail: florina.tuluca@gg.unibuc.ro

Accepted 2015 December 11. Received 2015 December 10; in original form 2015 June 16

SUMMARY

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016


Within a slab experiencing present-day lateral break-off, a particular type of earthquakes is
expected to cluster at the detachment horizon tip: namely, events generated by reverse faulting,
with the approximately horizontal compression involved acting along the strike of the slab.
Such a cluster of moderate magnitude earthquakes (4.7 ≤ mb ≤ 5.0) was identified in this study
at the 160–175 km depth range of the Vrancea seismogenic body, in the Southeast Carpathians
mountains collision environment. The corresponding cluster epicentres were systematically
positioned at the boundary between a region being subject (cf. published GPS records), to
present-day upward movements, and another one that underwent present-day subsidence.
Such an overall setting seems to suggest that a lateral break-off is currently developing at
the indicated depth within the Vrancea slab, leading to topographic uplift above the already
detached slab section, and to enhanced subsidence above the section to which the gravitational
slab pull was being transferred. In addition, by taking into account some systematic time
correspondence which we documented between moderate magnitude events of the 160–175 km
depth cluster and the subsequent strong Vrancea shocks (Mw ≥ 6.9), it appears that the latter,
although occurring at much shallower depths (roughly, in the 80–140 km range), were also
controlled by the break-off progress.
Key words: Seismicity and tectonics; Statistical seismology; Neotectonics; Kinematics of
crustal and mantle deformation; Europe.

domain, while simultaneously becoming more concentrated in the


1 I N T RO D U C T I O N
section that was still intact. Consequently, the slab section situated
Recent advances achieved in terms of 3-D numerical simulation above the detachment became subject to a reverse, upward directed
techniques have fostered a renewed interest in the issue of tectonic movement, while the continuous part of the slab kept on sinking
slabs undergoing lateral breaks-off processes (Burkett & Billen under its own negative buoyancy (Van Hunen & Allen 2011; Li
2010; Van Hunen & Allen 2011; Capitanio & Replumaz 2013; Li et al. 2013; Bottrill et al. 2014; Duretz et al. 2014).
et al. 2013; Bottrill et al. 2014; Duretz et al. 2014; Von Tscharner In contrast with the elaborate break-off scenarios proposed
et al. 2014). Specific behaviours having been originally postulated by the above-indicated numerical models, only very few studies
by earlier, pioneering works (e.g. Wortel & Spakman 2000), man- have so far attempted to diagnose—by relying on stress regime
aged now to be reproduced by means of computer simulations. A specificities suggested by the earthquakes focal mechanisms space
particular interest has been devoted to settings in which viscous distribution—actual settings that presently experienced lateral slab-
necking was initiated at one edge of the slab, after which the tearing detachment: Yoshioka & Wortel (1995) have considered in this
process migrated along the slab strike (Fig. 1). The horizontal con- respect the New Hebrides island arc region, Lister et al. (2008)
striction accordingly undergone by the slab at the tearing horizon have focused on the Hindu Kush, while Sandiford (2008) has ad-
tip corresponded to a localized shearing process, which involved dressed the Damar zone in the Banda Sea. Another particular setting
(Yoshioka & Wortel 1995) horizontal in-plane compression, ac- for which several authors (e.g. Wortel & Spakman 2000; Wenzel
companied by dip-parallel slab extension—hence a stress regime et al. 2002; Bonjer et al. 2008; Lister et al. 2008; Burkett &
liable to generate thrust fault earthquakes. Billen 2010) have hypothesized that seismicity patterns could be
As the tear progressively migrated laterally, the gravitational ver- indicative of a presently developing slab break-off is Vrancea re-
tical pull diminished in the slab section situated above the detached gion, at the Southeast Carpathians mountains bend (Fig. 2). So far

864 
C The Authors 2016. Published by Oxford University Press on behalf of The Royal Astronomical Society.
Lateral detachment within the Vrancea slab 865

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016


Figure 1. Main processes inferred to occur in association with an ongoing lateral slab detachment. The horizontal constriction undergone by the slab at the
tearing horizon tip is the result of a localized shearing process, which involves horizontal in-plane compression (orange arrow) accompanied by dip-parallel
slab extension (yellow arrows)—hence a stress regime liable to generate thrust fault earthquakes.

however, no detailed analysis of the Vrancea seismicity has ad- the associated deformation regime, by making use of several fo-
dressed specific signatures which could be ascribed to a slab cal mechanism catalogues which became available in the recent
detachment process. years (for instance Radulian et al. 2002; Sandu & Zaicenco 2008;
Consequently, this paper considered a detailed space–time dis- the European-Mediterranean Regional Centroid Moment Ten-
tribution of the Vrancea intermediate-depth earthquakes and of sor Catalogue—http://www.bo.ingv.it/RCMT/yearly_files.html; the

Figure 2. General seismotectonic setting of the considered area (surface geology after Săndulescu 1984). The index map illustrates the study area location
on the territory of Romania; VR refers to the boundary (solid oval) of the surface projection of the intermediate-depth seismicity Vancea region. Dots on the
main map represent the epicentres of the intermediate-depth (>60 km) Vrancea earthquakes occurred during the 1965–2011 time interval and for which both
the ISC and the PDE catalogues indicated mb ≥ 4.7. The colours of dots distinguish—as specified by the legend—the four groups of earthquakes identified,
cf. Section 3.2, within the seismogenic volume. Beach balls (lower hemisphere projections, with white quadrants indicating compression) illustrate the focal
mechanism solutions provided by the Global Centroid Moment Tensor Catalogue for the strongest Vrancea earthquakes (Mw ≥ 6.9) occurred during the
concerned period. The purple line is the trace of the vertical cross-section shown in Fig. 4, with numbers indicating the distance (in km) along the profile. The
green rectangle indicates the boundaries of the digital elevation model illustrated in Fig. 8.
866 H. Mitrofan et al.

catalogue of the Swiss Seismological Service—http://www. is the reverse faulting regime, with T-axes displaying a steep, to
seismo.ethz.ch/prod/tensors/index_EN). near-vertical position (e.g. Bala et al. 2003; Hurukawa et al. 2005).
That circumstance has been unanimously interpreted (e.g. Radulian
et al. 2000; Wortel & Spakman 2000; Sperner et al. 2001; Wenzel
et al. 2002; Cloetingh et al. 2004; Radulian 2014) as a signature of
2 SEISMOTECTONIC SETTING
gravitational stretching of the slab in the vertical direction. There
Positioned in a fully intracontinental setting within Europe, Vrancea is another systematic feature, which concerns the sub-horizontal P-
region nonetheless represents a major source of seismic hazard: axes of the four most recent strong (Mw ≥ 6.9) Vrancea earthquakes
earthquakes occur frequently in an intermediate-depth range (60– (Fig. 2): they all struck NW–SE (e.g. Oncescu & Bonjer 1997), that
180 km), with several major shocks (magnitudes in excess of 7) is, the compression was approximately perpendicular to the inferred
being recorded each century (Ismail-Zadeh et al. 2012). The seis- direction of the Vrancea slab. By additionally considering that also
mogenic body surface projection is highly confined (25 × 70 km many (yet not all) moderate magnitude earthquakes of the Vrancea
lateral extent) and it roughly corresponds to a present-day colli- area obeyed to the same stress pattern, Cloetingh et al. (2004) pre-
sional boundary: between the East European and Moesian foreland sumed that not only vertical extension arising from gravitational
units on the one hand, and the Intra-Alpine plate on the other. That slab-pull was responsible for the Vrancea seismic activity, but also
boundary is overlain (Fig. 2) by a Tertiary age fold-and-thrust belt mechanical interaction which possibly operated between the lower
that belongs to the Eastern Carpathians mountains. The Miocene and the overriding plates that collided in that region.
emplacement of the corresponding nappe pile has been the re- Elucidating such controversial issues involved by the currently
sult of an approximately westward-directed oceanic subduction, proposed conceptual models requires, among others, a detailed anal-

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016


which further evolved by continental collision, about 10 Ma ago. ysis of the space–time distribution of the Vrancea intermediate-
In that late stage, the subducted plate might have undergone (as depth earthquakes and of the correspondingly suggested stress
discussed, for instance, in Wortel & Spakman 2000; Sperner et al. pattern evolutions.
2001; Ismail-Zadeh et al. 2012) lateral tearing along the Carpathi-
ans orogen strike. The latter process appears to be consistent with
an SE-directed time-migration displayed by the eruptive activity
3 I N V E S T I G AT I O N A P P R O A C H
recorded in a volcanic chain developed in the backarc region. In
explaining that progressive migration of the Neogene volcanic ac-
3.1 The utilized earthquakes database
tivity along the East Carpathians arc, from NW to SE, Mason et al.
(1998), Wenzel et al. (2002), and Seghedi et al. (2004) admitted We began by compiling for the Vrancea intermediate-depth earth-
an oblique (diachronous) type of collision. It is worth emphasizing quakes a dedicated catalogue, which benefited of basic information
that specifically this kind of collisional geometry has been shown by (time of occurrence, hypocentre location, magnitude) substanti-
3-D numerical modelling (Bottrill et al. 2014) to be able to induce ated by more than one independent—and mutually corroborating—
lateral slab detachment. There has been recently some debate (e.g. sources. Accordingly, our primary assumption was that the most sig-
Seghedi et al. 2011; Ismail-Zadeh et al. 2012) on the hypothesis nificant earthquakes having occurred at intermediate depths within
of a slab detachment having propagated along the entire length of the Vrancea slab, were those for which both the International Seis-
the Eastern Carpathians range; such a controversy still does not mological Centre (ISC) and the Preliminary Determination of Epi-
preclude the possibility that currently, lateral tearing could be de- centres (PDE) catalogues (the latter being devised by the USCGS-
veloping just within a slab fragment preserved in Vrancea area (as NEIS-NEIC), indicated body-wave magnitudes (mb ) larger than, or
suggested by Wortel & Spakman 2000; Hackney et al. 2002; Bonjer at least equal to a prescribed threshold. This resulted in defining a
et al. 2008). new, composite catalogue, in which for each earthquake, the lowest
It has been determined in fact that the domain of the Vrancea of the two distinct mb values separately provided by the ISC and
intermediate-depth earthquakes occupied part of an SW–NE ori- PDE databases was ascribed.
ented high-velocity subcrustal body, which likely represented a relic The adopted procedure also required the identification of a
piece of subducted lithosphere. Out of the various seismic tomog- corresponding cut-off magnitude. For this purpose, a frequency-
raphy procedures which had outlined that structure, some (Wortel magnitude diagram was constructed (Fig. 3), by considering all the
& Spakman 2000; Piromallo & Morelli 2003; Martin et al. 2006) intermediate-depth (>60 km) Vrancea earthquakes occurred dur-
managed to establish that it extended as far as 350–370 km depth. ing the 1965–2011 time interval and for which both the ISC and
Others (Tondi et al. 2009; Koulakov et al. 2010) only utilized local the PDE catalogues indicated mb ≥ 4.5 (this magnitude threshold
earthquakes, and hence, were not able to image the high-velocity being selected as a value ‘to start with’). The opening year of the
body beyond 180 km depth. Yet none of the above-indicated to- chosen time interval, 1965, was the earliest when both considered
mographic studies succeeded to unambiguously highlight a region catalogues had reported Vrancea events with mb ≥ 4.5; while our
where the conjectured slab was discontinuous (as it had been es- analysis closing year, 2011, was the most recent one for which
tablished, for instance, for the Ionian slab in southern Italy—Neri seismological data reviewed by the ISC were currently available.
et al. 2009; Calò et al. 2012; Presti et al. 2013). Consequently, while The resulting frequency-magnitude distribution pattern (Fig. 3)
the Vrancea slab break-off model is nowadays rather popular, no appears to be quite similar to that which already Trifu & Radulian
consensus actually exists on the corresponding tearing depth: in the (1989, 1994) and, more recently, Telesca et al. (2012) had illustrated
40–70 km range—according to Mason et al. (1998), Wenzel et al. for the Vrancea population of intermediate-depth earthquakes. A
(2002), and Bonjer et al. (2008); around 100 km—according to significant deviation from a simple exponential relation is notable
Bokelmann & Rodler (2014); below 200 km—according to Wortel for strong shocks (mb > 5.5 in the case of the presently involved, ad
& Spakman (2000) and Martin et al. (2006). hoc catalogue), those large events occurring more frequently than
In terms of recorded focal mechanisms, a frequently noted, persis- predicted by a Gutenberg-Richter distribution. In order to explain
tent characteristic of the Vrancea intermediate-depth earthquakes that pattern, Trifu & Radulian (1994) conjectured that different
Lateral detachment within the Vrancea slab 867

lower mb value, the overall outcome was that for larger magnitudes,
it was the mb value reported by the ISC which we selected; while
for lower magnitudes, it was probable to also select USCGS-NEIS-
NEIC values—when they were smaller than those provided by the
ISC: it thus finally resulted, as compared to an ‘ISC-only’ cata-
logue, a bias toward an increased frequency of occurrence of the
lower magnitudes, this circumstance being ultimately expressed by
a steepening of the b slope. A somehow similar situation has been
illustrated also by Prieto et al. (2012) in their approach of determin-
ing the b slope for the Bucaramanga (Colombia) intermediate-depth
seismicity nest (Bucaramanga incidentally being, besides the Hindu
Kush, the only recognized analogue of the Vrancea intermediate-
depth seismicity setting): in order to alleviate discrepancies noted
between ISC records (which indicated a b slope of −1.35), and a
local catalogue (displaying a b slope of −1.05), Prieto et al. (2012)
have devised, from the two databases, a composite catalogue, which
provided a b slope of −1.60. (See also, for other possible sources
of variability in the b slope estimates, Godano et al. 2014 and ref-
erences therein.)

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016


Further elaborating on these topics is, however, beyond the objec-
tives of this study. Our specific purpose was just to detect a cut-off
value down to which the regression illustrated in Fig. 3 indicated
compliance to a Gutenberg–Richter distribution. Accordingly, it re-
sulted that the selection of events which we had initially considered
was complete only down to mb = 4.7.
Fig. 4 illustrates how the hypocentres of the ultimately selected
earthquakes (those having mb ≥ 4.7) are positioned in a vertical
cross-section traced along the conjectured strike (N50 ◦ E) of the
Figure 3. Frequency–magnitude diagram of intermediate-depth (>60 km) Vrancea slab. The two distinct images, (a) and (b), correspond to
Vrancea earthquakes occurred during the 1965–2011 interval. There are
hypocentre localizations indicated by the ISC and the PDE cata-
included all the events for which both the ISC and the PDE catalogues
indicated mb ≥ 4.5, to each earthquake being ascribed the lowest of the
logues, respectively.
two distinct mb values separately provided by each of the two indicated Yet only the ISC catalogue also specifies errors associated with
catalogues. the computed hypocentre depths: accordingly, as shown by Fig. 5,
during the first 30 yr of the investigation period (1965–1994), the
‘families’ of earthquakes coexisted within the same seismogenic depth resolution underwent no improvements, most errors ranging
body, a distinct genetic mechanism (crack-like failure/asperity-like between 1.5 and 4 km (still with a maximum at 5.1 km); for events
failure/percolation) being ascribed to each such ‘family’. Alterna- occurred during the 1998–2002 interval, the errors range narrowed,
tively, Godano et al. (2014) have shown that a similar distribution down to 1.5–2.5 km; while for the 2004–2009 earthquakes, a signif-
pattern was also characteristic to many other seismic regions around icant improvement became obvious (errors fell just in the 1–1.5 km
the world (California, Japan, Greece, Italy, Switzerland), and that range).
it was possible to successfully model that pattern by assuming that
the b slope had a roughly Gaussian distribution, controlled by the b
parameter spatial and temporal variability. 3.2 Location characteristics of the selected earthquakes
On the other hand, it appears that the b slope determined in hypocentres
the case of the regression illustrated in Fig. 3 is rather unusually
Irrespective of which catalogue is considered, a similar pattern of
steep (−1.53). Previous investigators of the Vrancea intermediate-
structures can be distinguished in both Figs 4(a) and (b): each such
depth seismicity (Bazacliu & Radulian 1999; Ardeleanu &
structure has the appearance of a cluster, within which distances
Bazacliu 2005; Prieto et al. 2012) have also determined (by us-
between adjacent hypocentres are, usually, smaller than 10 kilome-
ing, of course, various time intervals and magnitude databases—
tres; while the distance (essentially along the vertical) between two
that of the ISC included) b slopes steeper than −1; but the latter
adjacent clusters is significantly larger (of the order of 20 km). The
values fell, nonetheless, mostly within the range −1.1 to −1.35.
two most prominent such clusters occur within the cross-section
Therefore, it seems that responsible for additionally increasing the
central part: they include hypocentres that form a pair of distinct
b slope could be the very approach which we had adopted for
lineaments, both of which display an obvious dip from NE to SW:
compiling our earthquakes-database: specifically, by considering a
worldwide earthquakes-database, Scordilis (2006) has outlined (his (1) the ‘Upper diagonal lineament’ extends between 80 and
fig. 6) that for events stronger than M ∼ 5, the mb values reported by 160 km depth and it includes, besides moderate magnitude earth-
the USCGS-NEIS-NEIC were, quite systematically, larger than the quakes, also all three major events (mb > 6) of the analysed time
corresponding mb values reported by the ISC; alternatively, toward period (throughout our paper, those particular three events are re-
smaller magnitudes, it seemed that positive and negative deviations ferred to, most of the time, in terms of the mb magnitudes provided
between the two catalogues were more or less equally distributed. by the ISC or the PDE catalogues, while in certain instances they are
Since a similar behaviour occurred also in the case of the Vrancea specified with the corresponding Mw magnitude values indicated by
earthquakes, and because our approach was to always retain the Oncescu & Bonjer 1997);
868 H. Mitrofan et al.

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016


Figure 4. Vertical cross-section along the inferred strike (N50◦ E) of the Vrancea slab (the transect position is indicated in Fig. 2). The hypocentres of the
intermediate-depth (>60 km) Vrancea earthquakes which occurred during the 1965–2011 time interval and for which both the ISC and the PDE catalogues
indicated mb ≥ 4.7 are projected perpendicular to the plane of the cross-section. The two distinct images, (a) and (b), correspond to hypocentre localizations
indicated by the ISC and the PDE catalogues respectively. The distance of each hypocentre to the vertical plane of the profile is no larger than 11.2 km for the
ISC catalogue and 9.0 km for the PDE catalogue. Hypocentres seem to be distributed—irrespective of which catalogue is considered—into four distinct groups,
as specified by labels. There are highlighted the major events (mb > 6) of the concerned period, all three of them being located within the ‘Upper diagonal
lineament’ and being a result of thrust faulting induced by compression at high angle to the slab trend. Additional distinctions in terms of focal mechanism types
are made for the moderate magnitude events, ‘inconclusive’ referring to fault plane solutions that are drastically different between one providing catalogue and
another.

(Fig. 4), and a similarly ‘slanting’ region of minimum released seis-


mic energy, which separates two alignments of maximum energy
release, outlined by Lăzărescu et al. (1990) for about the same
transect along Vrancea area. Given that the latter authors’ analysis
covered a different time period (spanning between the years 1900
and 1989), and utilized a different Vrancea earthquakes catalogue
(that included events with Ms ≥ 4.0, with the corresponding magni-
tude estimates being provided by the particular references cited in
the considered paper), the indicated similarity suggests that struc-
tures which may be inferred from Fig. 4 are not an artefact of the
data processing approach. Such an overall setting might be, on the
other hand, consistent with the hypothesis discussed in Radulian
(2014) about the Vrancea strong shocks nucleating, repeatedly, on
just a few weakness surfaces which pre-existed within the slab.
Although not forming an actual cluster, two isolated hypocentres
situated far above the ‘Upper diagonal lineament’ (at around 70 km
depth) have nonetheless been designated as outlining an ‘Uppermost
Figure 5. Hypocentre depth errors reported by the ISC catalogue, plotted as group’.
a function of the computed depth of the hypocentres. Symbols discriminate Finally, in the bottom section of the seismogenic body, below
between periods of increasingly improved depth resolution. 160 km depth, we observe the ‘Lowermost group’ of events. As
compared to either the ‘Upper’, or the ‘Lower’ diagonal lineaments,
(2) the ‘Lower diagonal lineament’ extends between 120 and this ‘Lowermost group’ was far less seismically active, while at the
160 km depth and it exhibits the largest abundance of events, yet all same time its earthquake activity seems to display—in comparison
of them of moderate magnitude. to the ‘Upper diagonal lineament’—a notable time relation, in which
events of the ‘Lowermost Group’ seem to precede the main shocks
There is a notable coincidence between the about 20 km thick
of the ‘Upper diagonal lineament’ by a 3–4 yr time interval (Fig. 6).
‘aseismic’ domain which separates these two ‘diagonal lineaments’
Lateral detachment within the Vrancea slab 869

Swiss Seismological
Table 1. The Vrancea intermediate-depth earthquakes ascribed to the ‘Lowermost group’. That group includes the earthquakes with hypocentres located below 160 km depth (according to the 1965–2011 records

TF—thrust faulting. All TF solutions indicate along-strike compression (P-axes strike within ±25◦ of the inferred orientation of the slab, N50◦ E; with one exception—indicated with italic fonts—for which the
of the ISC and the PDE catalogues) and which comply with the adopted completeness criterion (i.e. an mb value of at least 4.7 indicated by each of the two concerned catalogues—see Section 3.1 for details).

Service

NF




Centroid Moment Tensor Catalogue
European-Mediterranean Regional

NF




from focal mechanism solutions independently provided by the indicated references
Stress regime derived (according to the criteria of Zoback 1992)

Global Centroid Moment


Tensor Catalogue
Figure 6. Distribution of the time occurrence of the considered Vrancea

TF

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016





included also in the Earthquake Mechanisms of European Area (EMMA) database (http://www.emsc-csem.org/Earthquake/emma.php).
intermediate-depth earthquakes, as a function of their location in one of
the four distinct subregions defined within the slab. Symbols as specified

NF—normal faulting. All NF solutions indicate along-strike tension (T-axes strike within ±25◦ of the inferred orientation of the slab, N50◦ E).
in Fig. 4. Arrows suggest that a particular, common evolution might be
responsible both for a major earthquake, and for the ‘Lowermost group’
moderate event (mb < 5) which preceded it by a time interval ranging

Sandu & Zaicenco


between 3 and 4 yr.

(2008)

TF
TF

TF
TF

It was the latter characteristic (which would be discussed in
more detail below) that made us focus primarily on the ‘Lower-
most group’. In addition, the determined focal mechanisms of this

Radulian et al.
cluster of events are, systematically, similar between all providing

(2002)
catalogues (Table 1); in contrast, for each of the other previously

TF


defined groups of events, various catalogues provided fault plane
solutions which on many occasions were, for one and the same mod-
erate magnitude earthquake, very different (see Tables S1–S3 of the Mostrioukov & Petrov
Supporting Information), and which were referred to as: ‘inconclu-
sive fault plane solutions’ (Figs 4 and 6). We estimated consequently
(1994)

TF
TF
TF

TF

that out of the entire set of moderate magnitude events which we had
considered, those of the ‘Lowermost group’ were the most promis-
ing for attempting to derive appropriate interpretations in terms of
slab deformation regime.
Oncescu (1987)a

TF
TF
TF


3.3 Particular patterns outlined in the ‘Lowermost group’


P-axis strike is within ±45◦ of the inferred orientation of the slab).

seismic activity
As illustrated by Table 1 and Figs 4 and 6, most events that belonged
5.0
4.8
4.9

4.7
PDE

5.0
Magnitude mb

to the ‘Lowermost group’ were due to thrust-faulting, with the cor-


responding P-axes striking, in all cases, SW-NE (i.e. approximately
4.9
4.9
4.9

4.8
ISC

5.0

parallel to the inferred slab orientation—Fig. 7). This characteris-


tic alignment of the P-axes in the plane of the slab, together with
164.0
174.0

163.3
168.9
PDE

162.5

the T-axes nearly vertical positions (which essentially indicated dip-


parallel slab stretching) could be due to an ongoing process of lateral
Depth (km)

detachment—similar to that which had been schematically depicted


2.0
Err (km)
2.2

2.5
5.1
1.7

in Fig. 1.
ISC

On the other hand, there can be noted that the period 1977–
1990, which included the most recent strong (Mw ≥ 6.9) Vrancea
160.8
171.1

160.4
165.9
172.1

earthquakes, is partially superimposed (Fig. 6) on the time interval


1973–1987, when in the ‘Lowermost group’ there had occurred all


Time of occurrence

23 Oct 1973 10:51

the moderate magnitude events due to thrust faulting. Moreover,


2 Oct 1978 20:29

30 Nov 2002 8:16


21 Mar 1987 3:01
25 Jan 1983 7:35

each of the large Vrancea earthquakes of the considered period


a Solutions

(1977 March 4, Mw 7.4; 1986 August 30, Mw 7.1; 1990 May 30,
Mw 6.9) had been preceded, within a time lag that ranged strictly
between 3 and 4 yr (Table 2) by a ‘Lowermost group’ thrust fault
870 H. Mitrofan et al.

one-year long time interval could be an appropriate choice for the


cells dimension. Our approach appears to be in accordance with a
relative uniformity which had been previously recognized (Trifu &
Radulian 1994; Bazacliu & Radulian 1999) in the time distribution
of the Vrancea intermediate-depth earthquakes.
Further details on the corresponding probability computations
are provided in the Appendix. The accordingly obtained results
suggested that the correlation noted between the LGTF earthquakes
and the subsequent major Vrancea shocks was, most probably, not
accidental.

4 DISCUSSION
For addressing the possible origin of earthquakes in the ‘Lowermost
group’, a broader reference framework has been considered by re-
lying on independent observations that seemed to point towards a
similar, coherent scenario.
Figure 7. Lower hemisphere projection of the P and T axes (a and b panels

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016


respectively) determined for the thrust fault earthquakes of the ‘Lowermost
group’. Each event is identified by the date of its occurrence and by a specific 4.1 Present-day vertical displacements
colour, while symbols discriminate between solutions provided by different The ground vertical displacements map derived from GPS measure-
authors (as indicated by references). Dashed lines mark the ±25◦ deviation
ments conducted in that region by Schmitt et al. (2007) displays
from the inferred slab strike, thus helping to visualize that the determined
P-axes orientations are mostly concordant with the slab trend.
a conspicuous pair of contrasting anomalies: uplift rates reached
about 4 mm yr−1 within the NE section of Vrancea epicentral
area, while adjacent to it, in the SW part of the epicentral area,
(hereafter LGTF) event. Only one (1978 October 2, mb 4.9) of the there were recorded subsidence rates of up to about 6 mm yr−1 . In
four LGTF events failed to obey (Fig. 6, Table 2) this regular pattern. spite of using fewer stations, an analogous map devised by Van der
And only one additional event in the ‘Lowermost group’ (2002 Hoeven et al. (2005) outlined a similar pattern, with the contrasting
November 30, mb 5.0) was a result of normal-faulting. vertical velocities being nevertheless encompassed—as a result of
It would be of interest to estimate what is the probability that a different data processing strategy—within a smaller range (from
chance alone was responsible for the previously mentioned rela- +1.5 to −2.2 mm yr−1 ). Still no matter which of the two above-
tionships outlined between the LGTF events and the strong (Mw ≥ indicated maps was considered, the epicentres cluster of the ‘Low-
6.9) Vrancea earthquakes. Such an analysis must rely on an appro- ermost group’ was always positioned in the region that separated
priate model of the regarded seismic events time distribution. Yet the present-day uplift domain from the domain of present-day sub-
recurrence patterns of the Vrancea intermediate-depth earthquakes, sidence (Fig. 8—where, for clarity, only the data of Schmitt et al.
although having been addressed by several detailed studies (Trifu 2007 were plotted). Such an overall image appears to remarkably
& Radulian 1994; Enescu et al. 2005, 2006, 2008; Byrdina et al. match certain particular behaviour schemes postulated by Wortel &
2006), are still a controversial issue. Spakman (2000; their figs 3 and 4): a slab break-off which propa-
We have therefore adopted in this respect, for the purpose of this gated laterally (in this particular case, from NE to SW), was generat-
analysis, a simplified approach: specifically, our assumption was ing earthquakes at the tip of the corresponding detachment horizon
that the observation period could be divided into N time intervals (in our case, at depths of 160–175 km, where faulting compliant
of equal duration (which we designated as ‘cells’), and that only with a lateral break-off setting—Fig. 1—was outlined by earth-
one of all considered earthquakes could occupy each such cell. quakes of the ‘Lowermost group’—Fig. 8); at the same time, in
This approach was suggested by a rough estimate of the average the local topography there were induced contrasting vertical move-
recurrence interval of the involved earthquakes: slightly less than ments (as illustrated by the Vrancea maps devised by Schmitt et al.
1 yr—resulted by dividing the 47 yr length of the observation period, 2007, and by Van der Hoeven et al. 2005)—namely uplift above
by the 48 events which had mb ≥ 4.7; at the same time, the three the already detached slab section, and subsidence above the slab
quasi-constant time-lags separating the major Vrancea earthquakes section to which the gravitational pull of the detached slab had been
from the preceding LGTF events systematically ranged between transferred. A somehow analogous, narrow cluster of intermediate-
3.19 and 3.60 yr (Table 2), supporting the assumption that about depth earthquakes, whose subhorizontal P-axes were aligned in the

Table 2. Correspondence between each of the LGTF earthquakes, and the subsequent major shock occurred in Vrancea region.
LGTF earthquake Subsequent major Vrancea earthquake

Time of occurrence Depth (km) Magnitude mb Time of occurrence Depth (km) Magnitude mb

ISC PDE ISC PDE ISC PDE ISC PDE Elapsed


◦ ◦
Err (km) Err (km) time (yr)
23 Oct 1973 10:51 171.1 2.2 174.0 4.9 4.9 4 Mar 1977 19:22 85.8 2.6 94.0 6.1 6.4 3.36
2 Oct 1978 20:29 160.8 2.0 164.0 4.9 5.0 30 Aug 1986 21:29 137.0 1.8 132.0 6.3 6.4 7.91
25 Jan 1983 7:35 160.4 2.5 163.3 4.9 4.8 30 Aug 1986 21:29 137.0 1.8 132.0 6.3 6.4 3.60
21 Mar 1987 3:01 165.9 5.1 168.9 4.8 4.7 30 May 1990 10:40 89.0 2.0 89.3 6.4 6.7 3.19
Lateral detachment within the Vrancea slab 871

the contact (Fig. 8) with the upper tectonic plate. Numerical mod-
elling (Yoshioka & Wortel 1995) indicates that ultimately, such fric-
tional forces that oppose to the additionally imposed vertical loading
may lead—within the slab section which is coupled to the overriding
plate—to thrust faulting along a plane that is parallel to the inter-
plate boundary (i.e. P-axes perpendicular to the slab strike). This
is in remarkable contrast with the thrust faulting recorded below—
next to the detachment horizon tip—where the almost horizontal
P-axes are roughly parallel to the slab strike (Yoshioka & Wortel
1995, their fig. 5).
There is a good consistency between the above described sce-
nario and the P-axes distinct orientations recorded in the case of the
considered Vrancea earthquakes (Fig. 8): approximately parallel
to the slab strike within the ‘Lowermost group’, and perpendic-
ular to the slab strike within the ‘Upper diagonal lineament’,
next to the contact with the overriding plate. It could also be ex-
plained in this way the previously discussed correlation noted be-
tween certain LGTF events induced by along-strike compression,
and the major (Mw ≥ 6.9) earthquakes which 3–4 yr afterwards oc-

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016


Figure 8. Interpretative sketch of the Vrancea slab lateral break-off set- curred several tens of kilometres above, within the ‘Upper diagonal
ting, proposing a comprehensive explanation for seismological observa- lineament’.
tions in correlation with the present-day topography vertical displacements. There is, moreover, another remarkable similarity between the
Schematic shapes are indicated for the Moesian plate, the Intra-Alpine plate, modelling results of Yoshioka & Wortel (1995), and the Vrancea
and the Carpathians accretionary wedge. Hypocentres are indicated, to- intermediate-depth seismicity patterns: the maximum shear stress
gether with the corresponding focal mechanisms, for: the strongest (Mw ≥ region which they identified within the slab section situated above
6.9) Vrancea earthquakes (cf. Global Centroid Moment Tensor Catalogue— the detachment horizon (their plate 5a) has an oblique position, be-
green quadrants indicating extension); the thrust fault earthquakes in the ing in this respect quite comparable to the ‘diagonal lineaments’ of
‘Lowermost group’ (cf. Oncescu 1987, and Radulian et al. 2002—red quad-
Vrancea hypocentres that were illustrated in our Figs 4 and 8. Yet
rants indicating extension); the normal fault earthquake in the ‘Lowermost
one cannot exclude either the possibility that the observed hypocen-
group’ (cf. Swiss Seismological Service—dark blue quadrants indicating
extension). The digital elevation model (DEM) boundaries are those shown tre lineaments represented in fact (e.g. Radulian 2014) inherited
in Fig. 2. Note that the DEM vertical scale is, as specified by the enclosed weakness surfaces, which pre-existed within the Vrancea slab.
colour code legend, exaggerated. Vertical arrows above the DEM indicate It is on the other hand expected that ensuing to future accumu-
uplift (red) and subsidence (dark blue) velocities, as derived from GPS lation of relevant seismic information, appropriate refinements will
records of Schmitt et al. (2007). For further explanations see Section 4. be brought to the above-discussed overall scheme, in order to under-
stand also the distinctive stress regime suggested by the remaining,
plane of the slab, has been interpreted by Sandiford (2008) to be the normal fault event which had been recorded in the ‘Lowermost
front of a laterally propagating rupture in the Banda Slab (Indone- group’.
sia); while the associated stress transmission within the rupturing
slab has been tentatively assumed by that author to be responsible for
the local topography evolution—in particular, for the anomalously 5 C O N C LU S I O N S
deep sea bathymetry recorded above the slab segment assumed to The specific procedure adopted in this study relied first of all on a
be still intact. simultaneous analysis of the seismological information which had
been provided both by the ISC and by the PDE catalogues. We were
consequently able to outline a set of large and moderate magnitude
4.2 Space–time distribution patterns in the Vrancea
(mb ≥ 4.7) earthquakes that likely were the most significant having
seismicity
occurred at intermediate depths (60–175 km) in Vrancea region
The 2-D thermo-mechanical model that Cloetingh et al. (2004) during the time interval 1965–2011.
had devised for Vrancea region has shown that the strong coupling The accordingly identified earthquakes appeared to be distributed
which existed between the subducted and the overriding plates was into four groups, which occupied distinct positions within the near-
expected to result in the so-called ‘two-sided’ collision (Faccenda vertical, narrow, SW–NE elongated Vrancea seismogenic body. In
et al. 2008; Li et al. 2011): a mantle lithosphere wedge from the the framework of this study, two of those groups resulted to be
overriding plate was dragged downward by the sinking lower plate, of particular interest: (i) the so-called ‘Upper diagonal lineament’,
to join in the vertical descent. The indicated model results have which was dipping from NE to SW, between 80 and 160 km depth,
been, in addition, invoked by its authors for tentatively ascribing and which included—besides several moderate magnitude events—
the Vrancea intermediate-depth earthquakes to interactions which also all three major Vrancea earthquakes (6.9 ≤ Mw ≤ 7.4) of the
operated at the contact between the lower (Moesian) plate and the considered period; (ii) the so-called ‘Lowermost group’, which oc-
dragged wedge of the overriding (Intra-Alpine) plate. curred as a more localized cluster in the 160–175 km depth range,
In this study, it is moreover conjectured that ensuing to the break- and its generally sparse seismic activity significantly increased dur-
off lateral propagation—which was, presumably, ‘boosted’ by each ing the 1973–1987 period, being thus partially superimposed on the
LGTF earthquake—the still continuous slab section situated above occurrence interval (1977–1990) of the previously indicated three
160 km depth experienced gravitational pull enhancements. The major Vrancea earthquakes: in fact, each of the latter three strong
corresponding increased traction exerted on that slab region is ex- shocks had been preceded, within a time lag that ranged strictly
pected to be strongly resisted by frictional forces which operate at between 3 and 4 yr, by a moderate magnitude earthquake belonging
872 H. Mitrofan et al.

to the ‘Lowermost group’. It appears to be quite unlikely (our esti- Bazacliu, O. & Radulian, M., 1999. Seismicity variations in depth and time
mates indicating a probability as small as 0.67 × 10−4 ) that such a in the Vrancea (Romania) subcrustal region, Nat. Hazards, 19, 165–177.
correspondence was accidental. Bokelmann, G. & Rodler, F.-A., 2014. Nature of the Vrancea seismic zone
The relationship between the Vrancea major earthquakes and the (Eastern Carpathians) – New constraints from dispersion of first-arriving
P-waves, Earth planet. Sci. Lett., 390, 59–68.
moderate magnitude events of the ‘Lowermost group’ is also re-
Bonjer, K.-P., Ionescu, C., Sokolov, V., Radulian, M., Grecu, B., Popa, M. &
flected by the corresponding focal mechanisms: while the strong
Popescu, E., 2008. Ground motion patterns of intermediate depth Vrancea
shocks were systematically a result of thrust faulting due to com- earthquakes: the October 27, 2004 event, in Harmonization of Seismic
pression ‘approximately perpendicular to the slab strike’, the pre- Hazard in Vrancea Zone, pp. 47–62, eds Zaicenco, A., Craifaleanu, I. &
ceding events of the ‘Lowermost group’ were always a result of Paskaleva, I., Springer.
thrust faulting due to compression ‘along the slab strike’. Such Bottrill, A.D., Van Hunen, J., Cuthbert, S.J., Brueckner, H.K. & Allen,
a space distribution of the principal stress axes is consistent with M.B., 2014. Plate rotation during continental collision and its relationship
the results of numerical modelling which had been conducted by with the exhumation of UHP metamorphic terranes: Application to the
Yoshioka & Wortel (1995) for lateral slab detachment settings. Norwegian Caledonides, Geochem. Geophys. Geosyst., 15, 1766–1782.
On the other hand, with respect to the Vrancea area topography Burkett, E.R. & Billen, M.I., 2010. Three-dimensionality of slab detach-
ment due to ridge-trench collision: laterally simultaneous boudinage
vertical displacements indicated by GPS records, the epicentres of
versus tear propagation, Geochem. Geophys. Geosyst., 11, Q11012,
the ‘Lowermost group’ events were systematically positioned at
doi:10.1029/2010GC003286.
the boundary that separated the Vrancea epicentral sector which Byrdina, S., Shebalin, P., Narteau, C. & Le Mouël, J.L., 2006. Temporal
experienced present-day uplift, from the other sector, which was properties of seismicity and largest earthquakes in SE Carpathians, Non-
subject to ongoing subsidence. That pattern is remarkably similar

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016


linear Process. Geophys., 13, 629–639.
to general conceptual models (e.g. Wortel & Spakman 2000) which Calò, M., Dorbath, C., Luzio, D., Rotolo, S.G. & D’Anna, G., 2012. Seis-
associate such differential vertical displacements to settings where mic velocity structures of southern Italy from tomographic imaging
the gravitational pull had been transferred from a slab section al- of the Ionian slab and petrological inferences, Geophys. J. Int., 191,
ready detached because of a lateral break-off progress, to the slab 751–764.
section that was still continuous. Capitanio, F.A. & Replumaz, A., 2013. Subduction and slab breakoff con-
trols on Asian indentation tectonics and Himalayan western syntaxis for-
By overall considering the above-discussed results, it appears that
mation, Geochem. Geophys. Geosyst., 14, 3515–3531.
nowadays, a slab detachment is propagating laterally, from NE to
Cloetingh, S.A.P.L., Burov, E., Matenco, L., Toussaint, G., Bertotti, G.,
the SW, in the Vrancea region. The tip of the corresponding tearing Andriessen, P.A.M., Wortel, M.J.R. & Spakman, W., 2004. Thermo-
horizon is positioned, presumably, close to the ‘Lowermost group’ mechanical controls on the mode of continental collision in the SE
of earthquakes—namely in the 160–175 km depth range. There is, Carpathians (Romania), Earth planet. Sci. Lett., 218, 57–76.
on the other hand, a systematic time relation between the Vrancea Duretz, T., Gerya, T.V. & Spakman, W., 2014. Slab detachment in laterally
major earthquakes occurrences and some of the preceding moder- varying subduction zones: 3-D numerical modeling, Geophys. Res. Lett.,
ate magnitude events of the ‘Lowermost group’. This circumstance 41, 1951–1956.
seems to suggest that the detachment progress also controlled the Enescu, B., Ito, K., Radulian, M., Popescu, E. & Bazacliu, O., 2005. Mul-
generation of the Vrancea strongest shocks—and this, in spite of tifractal and chaotic analysis of Vrancea (Romania) intermediate-depth
earthquakes: investigation of the temporal distribution of events, Pure
the fact that those major earthquakes occurred at much shallower
appl. Geophys., 162, 249–271.
depths (roughly, between 80 and 140 km). The accordingly required
Enescu, B., Ito, K. & Struzik, Z.R., 2006. Wavelet-based multiscale resolu-
long-range interactions could originate in the slab downward pull tion analysis of real and simulated time-series of earthquakes, Geophys.
enhancement (which had resulted from the break-off lateral prop- J. Int., 164, 63–74.
agation), being opposed by frictional forces that operated at the Enescu, B., Struzik, Z. & Kiyono, K., 2008. On the recurrence time of
contact between the lower tectonic plate, and a detached wedge of earthquakes: insight from Vrancea (Romania) intermediate-depth events,
the upper plate. Geophys. J. Int., 172, 395–404.
Faccenda, M., Gerya, T.V. & Chakraborty, S, 2008. Styles of post-subduction
collisional orogeny: influence of convergence velocity, crustal rheology
AC K N OW L E D G E M E N T S and radiogenic heat production, Lithos, 103, 257–287.
Godano, C., Lippiello, E. & de Arcangelis, L., 2014. Variability of the b value
We are indebted to our colleague Răzvan-Gabriel Popa for fruit- in the Gutenberg–Richter distribution, Geophys. J. Int., 199, 1765–1771.
ful discussions and for his perceptive remarks concerning the Hackney, R., Martin, M., Ismail-Zadeh, A., Sperner, B. & Ioane, D. &
manuscript. We also gratefully acknowledge the kind support that CALIXTO Working Group, 2002. The gravity effect of the subducted
Luminiţa Ardeleanu, Mihaela Popa, Maria Tumanian, Mircea Radu- slab beneath the Vrancea region, Romania, Geologica Carpathica, special
lian and Alin Tudorache provided us for enlarging the necessary issue in Proceedings of XVII Congress of Carpathian-Balkan Geological
database of scientific publications. Our manuscript underwent sig- Association, Bratislava, vol. 53, pp. 119–121.
nificant improvements as compared to its original version, ensuing Hurukawa, N., Popa, M. & Radulian, M., 2005. Vrancea intermediate-depth
earthquakes in Romania. Seismicity and mechanisms, J. Balkan geophys.
to insightful comments and suggestions provided by two anonymous
Soc., 8(Suppl. 1), 315–318.
reviewers, to whom we express our whole gratitude.
Ismail-Zadeh, A., Matenco, L., Radulian, M., Cloetingh, S. & Panza, G.,
2012. Geodynamics and intermediate-depth seismicity in Vrancea (the
south-eastern Carpathians): current state-of-the art, Tectonophysics, 530–
REFERENCES 531, 50–79.
Ardeleanu, L. & Bazacliu, O., 2005. Time evolution of seismic activity pa- Koulakov, I., Zaharia, B., Enescu, B., Radulian, M., Popa, M. & Parolai, S.,
rameters related to the occurrence of the strong earthquakes of Vrancea 2010. Delamination or slab detachment beneath Vrancea? New arguments
region, St. Cerc. Geofizică, 43, 41–51 (in Romanian with English ab- from local earthquake tomography, Geochem. Geophys. Geosyst., 11,
stract). Q03002, doi:10.1029/2009GC002811.
Bala, A., Radulian, M. & Popescu, E., 2003. Earthquakes distribution and Lăzărescu, V., Rădulescu, F., Vasilescu-Pelcea, G. & Smalbergher, V., 1990.
their focal mechanism in correlation with the active tectonic zones of Distribution of earthquake energy in the Vrancea Region, Rev. Roum.
Romania, J. Geodyn., 36, 129–145. Géophys., 34, 41–47.
Lateral detachment within the Vrancea slab 873

Li, Z.H., Xu, Z.Q. & Gerya, T.V., 2011. Flat versus steep subduction: Con- Telesca, L., Alcaz, V. & Sandu, I., 2012. Analysis the 1978–2008 crustal
trasting modes for the formation and exhumation of high- to ultrahigh- and sub-crustal earthquake catalog of Vrancea region, Nat. Hazards Earth
pressure rocks in continental collision zones, Earth planet. Sci. Lett., 301, Syst. Sci., 12, 1321–1325.
65–77. Tondi, R., Achauer, U., Landes, M., Davı́, R. & Besutiu, L., 2009. Unveiling
Li, Z.H., Xu, Z.Q., Gerya, T.V. & Burg, J.P., 2013. Collision of continen- seismic and density structure beneath the Vrancea seismogenic zone,
tal corner from 3-D numerical modeling, Earth planet. Sci. Lett., 380, Romania, J. geophys. Res., 114, B11307, doi:10.1029/2008JB005992.
98–111. Trifu, C.-I. & Radulian, M., 1989. Asperity distribution and percolation
Lister, G., Kennett, B., Richards, S. & Forster, M., 2008. Boudinage of a as fundamentals of an earthquake cycle, Phys. Earth planet. Inter., 58,
stretching slablet implicated in earthquakes beneath the Hindu Kush, Nat. 277–288.
Geosci., 1, 196–201. Trifu, C.-I. & Radulian, M., 1994. Dynamics of a seismic regime: Vrancea—
Martin, M. & Wenzel, F. & the CALIXTO working group, 2006. High- a case history, in Nonlinear Dynamics and Predictability of Geophysical
resolution teleseismic body wave tomography beneath SE-Romania—II. Phenomena, Vol. 83, pp. 43–53, eds Newman, W. I., Gabrielov, A. &
Imaging of a slab detachment scenario, Geophys. J. Int., 164, 579–595. Turcotte, D. L., Geophysical Monograph Series, American Geophysical
Mason, P.R.D., Seghedi, I., Szákacs, A. & Downes, H., 1998. Magmatic Union.
constraints on geodynamic models of subduction in the East Carpathians, Van der Hoeven, A.G.A. et al., 2005. Observation of present-day tectonic
Romania, Tectonophysics, 297, 157–176. motions in the Southeastern Carpathians: results of the ISES/CRC-461
Mostrioukov, A.O. & Petrov, V.A., 1994. Catalogue of focal mecha- GPS measurements, Earth planet. Sci. Lett., 239, 177–184.
nisms of earthquakes 1964–1990, in Materials of the World Data Cen- Van Hunen, J. & Allen, M.B., 2011. Continental collision and slab break-off:
ter, 87 pp., Moscow; also available at: http://www.ifz.ru/open_data/ a comparison of 3-D numerical models with observations, Earth planet.
focal-mechanisms/, last accessed 24 December 2015. Sci. Lett., 302, 27–37.

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016


Neri, G., Orecchio, B., Totaro, C., Falcone, G. & Presti, D., 2009. Subduction Von Tscharner, M., Schmalholz, S.M. & Duretz, T., 2014. Three-
beneath Southern Italy close the ending: results from seismic tomography, dimensional necking during viscous slab detachment, Geophys. Res. Lett.,
Seismol. Res. Lett., 80, 63–70. 41, 4194–4200.
Oncescu, M.C., 1987. On the stress tensor in Vrancea region, J. Geophys., Wenzel, F., Sperner, B., Lorenz, F. & Mocanu, V., 2002. Geodynamics, to-
62, 62–65. mographic images and seismicity of the Vrancea region (SE-Carpathians,
Oncescu, M.C. & Bonjer, K.P., 1997. A note on the depth recurrence and Romania), EGU Stephan Mueller Spec. Publ. Ser., 3, 95–104.
strain release of large Vrancea earthquakes, Tectonophysics, 272, 291– Wortel, M.J.R. & Spakman, W., 2000. Subduction and slab detachment in
302. the Mediterranean-Carpathian region, Science, 290, 1910–1917.
Piromallo, C. & Morelli, A., 2003. P wave tomography of the man- Yoshioka, S. & Wortel, M.J.R., 1995. Three-dimensional numerical mod-
tle under the Alpine-Mediterranean area, J. geophys. Res., 108, 2065, eling of detachment of subducted lithosphere, J. geophys. Res., 100,
doi:10.1029/2002JB001757. 20 223–20 244.
Presti, D., Billi, A., Orecchio, B., Totaro, C., Faccenna, C. & Neri, G., 2013. Zoback, M.L., 1992. First- and second-order patterns of stress in the
Earthquake focal mechanisms, seismogenic stress, and seismotectonics lithosphere: The World Stress Map Project, J. geophys. Res., 97,
of the Calabrian Arc, Italy, Tectonophysics, 602, 153–175. 11 703–11 728.
Prieto, G.A., Beroza, G.C., Barrett, S.A., López, G.A. & Florez, M., 2012.
Earthquake nests as natural laboratories for the study of intermediate-
depth earthquake mechanics, Tectonophysics, 570–571, 42–56.
S U P P O RT I N G I N F O R M AT I O N
Radulian, M., 2014. Mechanisms of earthquakes in Vrancea, in Encyclope-
dia of Earthquake Engineering, pp. 1–9, eds Beer, M., Kougioumtzoglou, Additional Supporting Information may be found in the online ver-
I.A., Patelli, E. & Siu-Kui Au, I., Springer. sion of this paper:
Radulian, M., Mândrescu, N., Panza, G.F., Popescu, E. & Utale, A., 2000.
Characterization of seismogenic zones of Romania, Pure appl. Geophys.,
157, 55–77. Table S1. Lower diagonal lineament.
Radulian, M., Popescu, E., Bala, A. & Utale, A., 2002. Catalog of fault plane Table S2. Upper diagonal lineament.
solutions for the earthquakes occurred on the Romanian territory, Rom. Table S3. Uppermost group.
J. Phys., 47, 663–685. (http://gji.oxfordjournals.org/lookup/suppl/doi:10.1093/gji/
Sandiford, M., 2008. Seismic moment release during slab rupture beneath ggv533/-/DC1).
the Banda Sea, Geophys. J. Int., 174, 659–671.
Sandu, I. & Zaicenco, A., 2008. Focal mechanism solutions for Vrancea
seismic area, in Harmonization of Seismic Hazard in Vrancea Zone, Please note: Oxford University Press is not responsible for the
pp. 17–46, eds Zaicenco, A., Craifaleanu, I. & Paskaleva, I., Springer. content or functionality of any supporting materials supplied by
Săndulescu, M., 1984. Geotectonics of Romania, 336 pp., Editura Tehnică the authors. Any queries (other than missing material) should be
(in Romanian). directed to the corresponding author for the paper.
Schmitt, G., Nuckelt, A., Knöpfler, A. & Marcu, C., 2007. Three dimen-
sional plate kinematics in Romania, in Proceedings of the International
Symposium on Strong Vrancea Earthquakes and Risk Mitigation,
A P P E N D I X : E S T I M AT I N G T H E
Bucharest, Romania, pp. 34–45.
Scordilis, E.M., 2006. Empirical global relations converting MS and mb to
P R O B A B I L I T Y T H AT T H E
moment magnitude, J. Seismol., 10, 225–236. C O R R E L AT I O N B E T W E E N L G T F
Seghedi, I. et al., 2004. Neogene–Quaternary magmatism and geodynamics E A RT H Q UA K E S A N D S U B S E Q U E N T
in the Carpathian–Pannonian region: a synthesis, Lithos, 72, 117–146. M A J O R V R A N C E A S H O C K S WA S
Seghedi, I., Maţenco, L., Downes, H., Mason, P.R.D., Szakács, A. & A C C I D E N TA L
Pécskay, Z., 2011. Tectonic significance of changes in post-subduction
Pliocene–Quaternary magmatism in the south east part of the Carpathian– As indicated in Section 3.3 of the main text, the observation period
Pannonian Region, Tectonophysics, 502, 146–157. has been divided into N time intervals of equal duration (which
Sperner, B., Lorenz, F.P., Bonjer, K.-P., Hettel, S., Müller, B. & Wenzel, F., we designated as ‘cells’), and we assumed that only one of all
2001. Slab break-off—abrupt cut or gradual detachment? New insights considered earthquakes could occupy each such cell. Next, we des-
from the Vrancea Region (SE Carpathians, Romania), Terra Nova, 13, ignated by G the particular number of cells which contained major
172–179. earthquakes (so that G < N), and assumed that those G cells always
874 H. Mitrofan et al.

occupied the same, fixed positions. Alternatively, each of the N − Then, the probability C, that out of the R number of LGTF earth-
G cells that were left could include any of the remaining moderate quakes having been generated during the entire series of N time
earthquakes. Out of the total number N − G of cells containing cells, G events are ‘anchored’ to a major earthquake, can be com-
moderate events, the particular number of cells with LGTF earth- puted as:
quakes was designated by R. In agreement with the setting actually
documented in Vrancea area, we assumed that R > G. F1 × F2 R! (N − 2G)!
C= = . (A4)
The particular instances requiring to be primarily investi- V (R − G)! (N − G)!
gated were those in which certain LGTF events were ‘anchored’
In the particular case that we considered, N = 48, R = 4 and
(Fig. A1) to their subsequent major earthquake: namely, the
G = 3, which leads to C = 2.82 × 10−4 .
instances in which each of the G cells occupied by a major earth-
Yet the actual observation data suggest that the number R −
quake was preceded—at a specified, fixed interval—by a cell be-
G, of the LGTF events other than those ‘anchored’ to the major
longing to the set of the R cells which included LGTF events.
earthquakes are in fact distributed only in-between these ‘anchored’
Subject to the above-indicated hypotheses, the number of such
events. Otherwise stated, between the first and the last LGTF events
possible arrangements—in which out of the R cells containing
‘anchored’ to the corresponding major earthquakes, there exists
LGTF events, G cells were ‘anchored’ to a corresponding major
(Fig. A1) a particular number of cells, D, and only those D cells
earthquake—resulted to be
‘are allowed to host’ the R − G number of ‘nonanchored’ LGTF
R! earthquakes.
F1 = . (A1) This additional constraint requires that it is also estimated the
(R − G)!

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016


possible amount of arrangements in which the R − G number of
Each such arrangement will occur on a number of F2 instances, ‘nonanchored’ LGTF events can be distributed within the D cells
corresponding to all the ways in which the remaining moderate (in fact, only within the D − (2G − 3) number of cells which are
earthquakes—other than the G number of LGTF events ‘anchored’ left available, considering that—in accordance with the Vrancea
to the major ones—can be distributed within the N − 2 G cells that area situation—the fixed interval specified between an ‘anchored’
are left available. The number of these instances is LGTF event and its subsequent major earthquake was assumed to
be no longer than the smallest interval existing between two such
F2 = (N − 2G)!. (A2) major earthquakes—Fig. A1). It is accordingly defined an additional
factor, F3 :
As for the total number of possible situations, V, it corresponds to (D − 2G + 3)!
the ensemble of the ways in which all the moderate earthquakes can F3 = , (A5)
(D − G − R + 3)!
occupy the N − G cells which are available to them. Accordingly,
V results to be: which represents the number of the above specified arrangements,
and which is needed for multiplying the number F1 , of the pos-
V = (N − G)!. (A3) sible arrangements in which out of the R cells containing LGTF

Figure A1. Illustration of the simplified scheme that we adopted for estimating the probability that chance alone was responsible for the correlations noted
between certain LGTF earthquakes and the subsequent major shocks. Cells represent conventional time intervals of equal duration (slightly less than 1 yr
long), each cell being assumed to include only one earthquake. The G dark grey cells are those assumed to contain one major event each, and to occupy,
consequently, fixed positions. Alternatively, the moderate events may occur at random in any of the remaining cells. Out of the total number of cells occupied
by moderate events, R cells contain LGTF earthquakes, and out of these R cells including LGTF earthquakes, G cells (light grey filling) are ‘anchored’ to the
subsequent cell occupied by a major shock (for clarity, the 9 additional cells preceding the illustrated series, as well as the 21 cells which followed after it, are
not represented in the figure, since they neither contain major events nor are they ‘anchored’ to major events). Symbols indicate—in correspondence with the
indicated time scale—the actual occurrence times of the Vrancea zone earthquakes of interest: diamonds—major events; triangles—LGTF events.
Lateral detachment within the Vrancea slab 875

events, G were ‘anchored’ to a corresponding major earthquake. For F1 × F  2 × F3


computing F1 eq. (A1) still holds in this case, and similarly, eq. (A3) C =
V
remains valid for computing V. Alternatively, eq. (A2) has to be al-
tered to: R! (N − G − R)! (D − 2G + 3)!
= . (A7)
(R − G)! (D − G − R + 3)! (N − G)!
F2 = (N − G − R)! (A6)
Given that for the considered case D = 13 (Fig. A1), it results
because now, the number of cells in which the remaining moderate C = 0.67 × 10−4 . The very low values obtained for C, and fur-
earthquakes can permute is only N − G − R. thermore for C , seem to suggest that the noted connection between
Overall, compliance to the additionally imposed constraint finally the LGTF earthquakes and the subsequent major Vrancea shocks
results in the following, new probability equation: might be nonrandom.

Downloaded from http://gji.oxfordjournals.org/ at New York University on March 17, 2016

You might also like