You are on page 1of 19

Tectonophysics 581 (2012) 144–162

Contents lists available at SciVerse ScienceDirect

Tectonophysics
journal homepage: www.elsevier.com/locate/tecto

Keys and pitfalls in mesoscale fault analysis and paleostress reconstructions, the use
of Angelier's methods
Jean-Claude Hippolyte a, b, c,⁎, Françoise Bergerat d, Mark B. Gordon e,
Olivier Bellier a, b, c, Nicolas Espurt a, b, c
a
Aix-Marseille Univ, CEREGE, UMR 7330, 13545 Aix en Provence cedex 4, France
b
CNRS, CEREGE, UMR 7330, 13545 Aix en Provence cedex 4, France
c
IRD, CEREGE, UMR 161, 13545 Aix en Provence cedex 4, France
d
ISTeP, UMR 7193 CNRS-UPMC, Université Pierre & Marie Curie, Case 117, 4 place Jussieu, 75252 Paris cedex 05, France
e
Shell International Exploration Production, Bellaire Technology Center, 3737 Bellaire Blvd., Houston, TX 77025, USA

a r t i c l e i n f o a b s t r a c t

Article history: Whereas most of the stress inversion methods using fault slip data only minimize the angle between the
Received 25 January 2011 measured striation and a computed shear stress to find the best fitting reduced stress tensor, Angelier
Received in revised form 31 December 2011 (1990) proposed an alternative method named INVD that also takes into account the relative shear stress
Accepted 11 January 2012
magnitude which allows the fault to move. Using artificial datasets and particular fault geometries we com-
Available online 20 January 2012
pare this method with one of the classical methods based on the minimization of the shear-slip angles (R4DT;
Keywords:
Angelier, 1984) and we show that in most cases the new method has improved the quality of the results. Fur-
Fault analysis thermore, as proposed by Angelier, we point out that the quality of the stress inversion primarily depends on
Stress inversion the quality of the field data. We give advice and warn about some pitfalls concerning determination of sense
Angelier of slip on fault planes, recognition of successive faulting events and their chronology, drawer (or wedge)
Paleostress faults, stress permutations, faults in vertical bedding. We also argue that, in case of tilted sequences, fault
Provence diagrams should not be presented without bedding planes. But we show that stress inversions, when realized
Italy with caution and with the correct method, can have much more applications than reconstructing stress fields,
like for determining: the paleo-horizontal, the nature and the sense of motion of large faults, the chronology
and age of large structures.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction They are generally based on a comparison between the measured


fault striation and a computed shear stress. In 1990, Angelier pro-
Stress inversions from fault slip measurements have now become posed a new method named INVD taking into account the relative
a common tool in tectonics. They are used for characterizing ancient value of the shear stress in addition to its direction. After 20 years
stress fields, and also in active tectonics (e.g., Angelier, 1994; Bellier of using the classical (R4DT) and the new (INVD) methods, we reflect
and Zoback, 1995; Shabanian et al., 2010) and can be used with earth- on their advantages and disadvantages. In this paper we compare
quake focal plane mechanisms to obtain the seismotectonic stress these methods for particular fault geometries that were not previ-
(e.g. Angelier, 2002; Carey-Gailhardis and Mercier, 1987; Gephart ously discussed. We wish to discuss the limitations of the two
and Forsyth, 1984). The theoretical aspects or validity of these types of methods to validate the choice of one or other according to
methods have also been discussed in some papers (e.g., Lisle and the geometry of the fault datasets.
Srivastava, 2004; Lisle et al., 1998; Orife and Lisle, 2006; Shan and The quality and reliability of paleostress analysis primarily depend
Fry, 2006; Shan et al., 2006; Twiss and Unruh, 1998). on the quality of field observation, which provides the input data for
Starting in the middle of the seventies, several methods have stress tensor calculations. Therefore, we provide recommendations
been proposed to invert for stress directions (e.g., Angelier, 1975, for the acquisition of reliable field observation necessary for fault
1979, 1984; Angelier and Goguel, 1979; Armijo and Cisternas, 1978; analysis and we underline some common errors often found in the lit-
Carey, 1976, 1979; Carey and Brunier, 1974; Etchecopar et al., 1981). erature. We point out that some fault geometries might be misunder-
stood and finally we illustrate, through new results, how paleostress
reconstruction might be useful in structural geology. We focus on
⁎ Corresponding author at: CEREGE (UMR 7330 CNRS), Université Aix-Marseille III,
BP 80, Europôle Méditerranéen de l'Arbois, 13545 Aix en Provence Cedex 4, France.
Angelier's methods because we have expertise in their utilization
Tel.: + 33 4 42 97 17 70, + 33 6 75 73 61 24; fax: + 33 4 42 97 16 58. and the Angelier's R4DT method is equivalent of most inversion
E-mail address: hippolyte@cerege.fr (J.-C. Hippolyte). methods. It should be noted, however, that the pitfalls discussed

0040-1951/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.tecto.2012.01.012
J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162 145

also apply to other methods (e.g. Carey, 1979; Etchecopar et al.,


1981).

2. Fault analysis

The quality of a paleostress reconstruction depends on the quality


of the field observations and slip-vector measurements. Jacques
Angelier encouraged students and colleagues to establish, as far as
possible, (i) the relative age succession of the analyzed structures,
and (ii) the reliability of the sense of fault slip. Accordingly, the “sup-
posed”, “probable”, and “certain” senses noted in the field are also in-
cluded in Angelier's software (Angelier, 1984, 1990). His approach of
paleostress computation is based on trial and error with the original
data including field observations (slip criteria, fault slip chronologies,
fault pattern, bedding attitude, size of fault, fault throw, synsedimen-
tary tectonics, etc.). This approach allows a more realistic paleostress
result than an approach only based on mathematical criteria. In this
section, we present some examples of small structures that are useful
in the field for analyzing faults but have rarely been described in pub-
lications, if at all.

2.1. Sense of fault slip

Determining the sense of slip on a fault is the first requirement for


making fault analysis. In sedimentary rocks, stratigraphic offset is of
course the most obvious criterion for the sense of displacement.
Nonetheless many faults record several episodes of movement, with
younger events often obliterating the structures related to older
ones. Hence, the observed offset might not be related to the preserved
striation. For example, normal faults are commonly reactivated with a
strike-slip motion, and the observed normal offset is no longer docu-
mented by corresponding transport lineations on the fault surface.
Faults with apparent normal stratigraphic offsets may also result
from strike-slip motions in a tilted sequence. To avoid these pitfalls,
Fig. 1. The R (Riedel) criterion of sense of movement on fault surface. On fault surfaces,
the sense of slip needs to be determined by observation of all struc- non-striated areas are generally interpreted as shelter zones, but in the case of R frac-
tural elements preserved on the fault surfaces. Of course, a compari- tures they face the movement of the missing block and the interpretation of the fault
son with stratigraphic offset remains necessary. movement would be contrary.
The classical criteria for shear sense determination were described
in many previous publications (e.g. Petit, 1987; Petit et al., 1983;
Vialon et al., 1976). However, we want to point out that the sense evolution and kinematics of which are matter of controversy (e.g.
of fault slip might be misinterpreted like in the case of the “Riedel” Argnani et al., 2009). Most geodynamic models suggest sinistral dis-
sense criterion (Fig. 1; Petit, 1976, 1987; Petit et al., 1983; Vialon et placement along the Mattinata fault since the Late Cretaceous (e.g.
al., 1976). In the Riedel-type experiments of shear (Riedel, 1929), sec- Schettino and Turco, 2011). Left-lateral slip is supported by the in-
ondary faults that penetrate the rock from the fault surface more or terpretation of Miocene calcarenites as pull-apart basin deposits be-
less obliquely are found to be synthetic (R, with an angle of about tween the two main segments of this fault (Fig. 2; e.g. Funiciello et
15°–30°) or antithetic (R′, with an angle of about 60°–75°) to the al., 1988). However, normal-slip indicators on a fault scarp bordering
main fault. By analogy, Petit (1976, 1987) named R (Riedel) criterion, this basin to the south, suggest that the Mattinata fault underwent
secondary synthetic shear fractures of R orientation found on fault several episodes of movement during its long history.
surfaces (Fig. 1). These fractures allow determination of the sense of Even if left-lateral motion probably occurred on this fault, as sug-
slip, because they dip in the direction of movement of the missing gested by the reconstruction of a NE-trending compression (e.g.
block. They are striated, and their intersection with the fault surface Hippolyte, 1992, Fig. 2 site 10), Argnani et al. (2009) point out that
may have a characteristic crescent shape. This “Riedel” criterion for this motion was much less than 2 km according to the size of the
determining the sense of slip is most common on faults cutting highly Miocene basin, which was probably not sufficient to form the ~200 m
competent rocks, such as granitoids or quartz rich sandstones, but wide damage zone present along the Mattinata fault. Likewise, we will
may occur on large-scale faults cutting less competent rock types show that our fault kinematics study reveals a complex history charac-
(e.g. limestones). They generally show a non-striated zone facing terized by a principal motion in a right-lateral sense (Fig. 2; Hippolyte,
the movement of the missing block. This last characteristic may lead 1992).
to the misinterpretation of the shear sense if they are mistaken for We measured striated fault surfaces in several sites along the
slickenside steps (Fig. 1). In order to correctly identify the nature of Mattinata fault and computed paleostresses. In the sites located out-
non-striated steps on fault surfaces it is therefore advisable to ascer- side the cataclasites of the fault damage zone (sites 10, 13, 9, 6, 12, 5
tain the presence or absence of striated shears underneath such steps. and 2, Fig. 2), we found mesoscale faults whose sense of slip is reli-
The failure to recognize the “Riedel sense indicator” can lead to ably determined with the stylolites and steps of calcite criteria. Most
misinterpretation of the sense of movement of large faults. For ex- of these fault sites are characterized by dextral slips on the E–W
ample, the E–W Mattinata fault (Fig. 2) is an important crustal struc- trending faults. They allowed computation of mainly (seven sites
ture with a long tectonic history. It displaces the Mesozoic platform out of eight) strike-slip state of stress with NW–SE compression,
carbonates of the Gargano promontory in southern Italy, the supporting a right-lateral motion of the Mattinata fault (Fig. 2).
146 J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162

Fig. 2. The use of the R (Riedel) criterion and of the paleostress reconstruction for determining the sense of slip of the Mattinata fault in the Gargano “spur”, Italy (coordinates:
41°43.2′ N 15°51.6′ E). A, B and C: Schmidt diagrams, lower hemisphere, with faults, bedding planes (dashed lines) and computed stress axes (INVD method, Angelier, 1990).
Five-branch star = σ1 (maximum principal stress axis); four-branch star = σ2 (intermediate principal stress axis); three-branch star = σ3 (minimum principal stress axis). D. Geo-
logical map of the Gargano peninsula with location of the fault sites and of the main Late Jurassic (Monte La Serra, San Giovani Rotondo, Monte Spigno Formations), and Cretaceous,
limestone facies (Servizio geologico d'Italia, 1965, 1969, 1976). The sense of movement of this ~ 150 km long fault system that extends into the Adriatic Sea is a matter of debate. We
measured mesoscale faults along its two main onshore segments and computed paleostresses (Hippolyte, 1992). Whereas most of the NW–SE faults of the Gargano peninsula are
normal faults resulting from NE–SW extension (σ3) (inset B), most of the mesoscale faults along the Mattinata fault system indicate strike-slip state of stress with NW-trending
compression (σ1) (inset C) suggesting a right lateral motion, which is confirmed by the senses of fault slips in the fault damage zone as indicated by the R criterion on fault surfaces
(photo).

We also measured fault planes directly in the cataclasites of the wide fault damage zone. A left-lateral motion of the Mattinata fault
fault zone at site 22 (Fig. 2) where large fault surfaces with senses is possible, but is only supported by one fault site in our study
of slip only determinable by the Riedel criterion occur. The E–W faults (Fig. 2, site 10) and it was less than 2 km (Argnani et al., 2009).
show right lateral slips. NW–SE dextral faults and N–S sinistral faults We have no chronology between these two events, but given that
can be interpreted as R and R′ shears according to the Riedel-type the Miocene rocks were affected by the NW–SE compression, at least
experiment (Riedel, 1929). Altogether these faults indicate a NW–SE part of the right lateral slip is Neogene. Anyway, the main right
shortening direction (Fig. 2 site 22) which confirms the dextral mo- lateral motion of the Mattinata fault, which is similar to the motion
tion of the Mattinata fault. However, if the Riedel criterion had not considered for the E–W offshore Tremiti fault located just north of
been recognized on the fault surfaces, its main slip could have been the Gargano peninsula (e.g. Doglioni et al., 1996), could be identified
interpreted as sinistral by analogy with the striation on one side of on the fault surfaces of the fault zone, only using the “Riedel” sense
asperities in a fault surface (Fig. 1). criterion.
Even if a sinistral movement on this fault is probable, as supported
by a NE–SW compression (Fig. 2, site 10; Hippolyte, 1992), the Matti- 2.2. Recognition of successive faulting events
nata fault zone shows that its main slip was right-lateral (Hippolyte,
1992). Indeed, this main dextral slip could be as large as 10 km as In case of polyphase tectonics testified by several slickenline di-
indicated by the apparent offset of late Jurassic rocks with different rections on fault planes, the separation of fault data subsets corre-
limestone facies (e.g. San Giovani Rotondo Limestone), as shown by sponding to different state of stress is feasible by fault categorization
the geological maps (Fig. 2; Servizio geologico d'Italia, 1965, 1969, (e.g. Célérier, 1995) or by computation only (e.g. Otsubo et al., 2006;
1976). This main right lateral motion can also explain the ~200 m Shan and Fry, 2005), but it can lead to significant mistakes (Sperner
J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162 147

and Zweigel, 2010), for example due to possible extreme values of the if they were traced after. This situation can be also recognized by the
tested stress ellipsoid shape factor. Angelier and Manoussis (1980) presence of stylolites on one side of the ridges of the corrugations
developed software for an automatic dataset separation using a dy- and/or fibrous calcite on the other side (Fig. 3). It follows that when
namic clustering approach. However, Angelier preferred the sepa- the steps of calcite of the second movement are aligned en-échelon
ration of data by using field observations when possible, with a along the direction of the first lineation the chronology of successive
computation approach allowing each fault to be validated accord- slip directions can also be ascertained (Fig. 3).
ing to its observation in the field. This approach is time consum-
ing but computation of data validated by field evidence produces
2.2.2. Fault offset chronology
more reliable results. However, chronologies of faulting events
Another method to establish the relative age relationship between
are not always clear, which can lead to mistakes. We present some
different fault generations is the cross-cutting relationship between
examples of chronological criteria from fault surfaces and fault geome-
faults, provided that the fault offset corresponds to the striation ob-
tries, which can reliably be used.
served on the fault plane (see some examples in Bergerat, 1985, 1987,
1994). We illustrate here the separation of two successive normal
2.2.1. Chronology of fault striations
faulting events from an outcrop at Puerto Rico (Fig. 4) where exten-
The chronology of superposed striations on a fault surface is
sional deformation is related to transtension along the strike-slip
often difficult to determine which can lead to mistakes. A com-
margin of the North-Caribbean plate (Hippolyte et al., 2005; Mann
mon cause of error is the observation of two generations of differ-
et al., 2005). Fig. 4 shows two fault families that can be separated
ently oriented fiber lines on superposed slices of fibrous calcite
based on cross-cutting relationships. Effectively, the E–W trending
infilling a fault plane (Sperner and Zweigel, 2010). The fiber
normal faults (family 1) are systematically offset in normal sense
lines generation covering the other is often considered as the
by the N–S trending normal faults (family 2). The separation of the
most recent, but its present exposition not necessarily establishes
two fault families in two fault diagrams and two state of stress is
a relative age criterion. More generally, slickenside chronologies
supported by their chronological relationship. This chronological
are ambiguous when the fault surface is smooth and when the
relationship is also supported by the observation that the first faults
two striations have the same intensity.
are syn-depositional of the early Miocene carbonates. These Miocene
Here we would like to point out that clear criteria of chronology
layers vary in thickness across the large E-trending faults, which are
of striations can be observed if the second movement was of lower
sealed by the upper layer of carbonate (Fig. 4). Notice also that the
magnitude than the first one (Fig. 3). When a first movement is
presence of oysters on the footwall of one of the first generation
large enough to create corrugations (grooves) in the fault plane,
faults, also attests to its syn-depositional activity.
and when it is followed by a different slip with a lower magnitude,
The separation of the two fault datasets allows characterizing an
the chronology can be justified by the fact that the second movement
early Miocene NNW–SSE syn-depositional extension, and a post-early
has not completely erased the ridges of the corrugations. On the
Miocene E–W extension which is in relation with the opening of
contrary, it is clear that the grooves would have erased the striation
the N-trending Mona rift between Puerto-Rico and Hispaniola
(Mann et al., 2002). If all the faults were used altogether to com-
pute a stress tensor one would have concluded that the extension
had been multidirectional because of the ratio the principal stress
differences Φ (e.g. Angelier, 1994) would be close to zero (see
next section).
The chronology of deformation events can also be deduced from
cross-cutting relationships between faults and bedding planes. This
is illustrated by an example from northern Turkey (Fig. 5). Normal
faults (shown in black) have offset late Cretaceous red limestone
by 50 cm to 1 m. These faults have however subsequently been also
tilted and offset by bedding parallel slip (in white on the photo-
graph) related to folding and strike-slip faulting as shown by fault
slip analysis (Fig. 5). Such cross-cutting relationship indicates that
compression followed extension. The outcrop scale fault chronology
can clarify the chronology of regional structures. In this example,
extension is related to the opening of the Black Sea during the Creta-
ceous, and compression is related to the formation of the Pontides
fold-and-thrust belt and the deformation along the North Anatolian
fault.

3. Paleostress reconstruction

Paleostress computation from fault slip measurements allows de-


termination of a “reduced stress tensor”: that is the orientation of the
three principal stress axes, the maximum (σ1), intermediate (σ2) and
minimum (σ3) principal stress axes, as well as the ratio of the princi-
pal stress differences Φ = (σ2 − σ3) / (σ1 − σ3), also called shape ratio
of the stress ellipsoid. However, one should keep in mind that the
ratio Φ is poorly constrained in the absence of oblique slip faults
(e.g. Angelier, 1994) because the shear stress does not change when
Fig. 3. Example of clear chronology of fault striation. If a striation has a lower magni-
tude than another which created corrugations (grooves), steps of calcite or stylolites
the Φ ratio changes where the fault sets are perfectly conjugate
on the corrugation sides indicate the subsequent fault movement. The photo illustrates (Angelier, 1994). Effectively, in the case of conjugate faults, the inter-
horizontal steps of calcites at the edges of vertical slickenlines. mediate stress axis is parallel to the fault intersection direction and
148 J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162

Fig. 4. Example of fault chronology from cross-cutting relationships near Aguada, Puerto Rico, Northern Caribbean plate boundary zone (coordinates: 18°21.5′ N 67°9.7′ W) (mod-
ified from Hippolyte et al., 2005). Stratigraphic layers in white. E–W normal faults, in black on the photo and on the fault diagram (NNW–SSE, extension) are cut and offset by N–S
normal faults (E–W extension) in gray. The cross-cutting chronology is confirmed by the observation that faults of the first family are syn-depositional (variation of thickness of the
Cibao early Miocene carbonates across the faults, and presence of oyster fossils on the uplifted footwall block).

the slip occurs in the plane defines by σ1 and σ3 whatever the relative episodes of faulting, and we look for oblique faults if we wish to con-
value of σ2. strain the Φ ratio.
In theory, for mathematical reasons, only four fault-slip mea-
surements are needed to compute the reduced stress tensor. 3.1. The two main types of paleostress inversion methods
Orife and Lisle (2006) however emphasized the danger of paleos-
tress determinations from less than eight faults. But for the reason Apart from the “right-dihedron method” (Angelier and Mechler,
mentioned above, the accuracy of the computed stress tensor de- 1977), that only provides a confidence ellipse containing two of the
pends not only on the number of fault slip data but also of the principal stress axes, all existing methods for stress computation are
orientation variability of the fault planes. Usually, when it is possible, based on the stress-shear relationship described by Bott (1959) and
we measure at least ten to twenty fault slip data for a single faulting by Wallace (1951). The problem of determining the stress tensor
event (note that too many faults could not clearly be displayed in knowing the direction and the sense of slip on numerous faults was
fault diagrams), and more if the rock formation underwent several first solved by Carey and Brunier (1974). Based on this work, various

Fig. 5. Example of chronology from cross-cutting relationships between faults and bedding-plane-parallel faults related to folding, late Cretaceous limestone, northern Turkey, near
Bafra (coordinates: 41°22.92′ N 35°48.76′ E).
J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162 149

developments and improvements have been proposed (e.g., Angelier, friction law or a value for the magnitude of the vector S (cf.
1975, 1984; Armijo and Cisternas, 1978; Carey, 1976; Carey- Angelier, 1990 for more details), one can see in Fig. 6, that the modu-
Gailhardis and Mercier, 1987; Delvaux and Sperner, 2003; Etchecopar lus of vector υ decreases as the shear stress modulus (τ) increases,
et al., 1981; Gephart and Forsyth, 1984; Hardcastle, 1989; Hardcastle which corresponds to a necessary requirement. Conversely, the mod-
and Hills, 1991; Lisle, 1988, 1989; Michael, 1984; Oncken, 1988; ulus of vector υ increases when the shear stress magnitude is small,
Reches, 1987; Simòn-Gòmez, 1986; Will and Powell, 1991). All even if the angle α is small (Fig. 6). The double significance of the
these computation methods were developed according to the basic adopted basic criterion should provide a more realistic solution com-
assumptions: (1) that the slip (indicated by striation) on each fault pared to classical methods. Anyway, it ensures that the acceptable av-
plane of a dataset has the direction and the sense of the resolved erage stress tensor solution does not result from fault planes with
shear stress on this plane; (2) that the slip on a given fault is not influ- anomalous low levels of relative shear stress.
enced by interaction with other faults, which is generally true for
small fault slips; and (3) that all fault slips correspond to a single
common stress tensor. The principle of these computation methods 3.2. Comparison and limitations of the two types of inversion methods
is to find the best fitting stress tensor to minimize a function of the
angle between the computed shear stress and the actual slip vector. To demonstrate the improvements of the INVD method, Angelier
To achieve this, most of these methods have an exploratory approach (1990) compared his new method with one of the inversion methods
and compare, through 3D or 4D search or iterative processes, numer- that minimize a function of the angle between the computed shear
ous tensors in order to find the minimum value of a sum of function. stress and the actual slip vector: his R4DT method (Angelier, 1984).
With the exception of Michael (1984), who simplified the inversion For this purpose he used homogeneous (single-phase deformation)
by the assumption that the magnitude of the tangential traction was fault datasets.
similar on the various planes, none of these methods consider the In practice, many fault sites recorded polyphase deformation and
magnitude of the shear stress on the faults. contain heterogeneous fault slips related to the successive stress
Angelier (1990), however, noticed that these methods based on field orientations. The computation of paleostress axes with faults
the minimization of the shear-slip angles give imprecise solution in originating from different states of stress would of course result in a
some cases. When a stress vector on a fault plane is close to the nor- composite stress tensor and give an incorrect result. Therefore one
mal of this plane, small variations in slip plane orientation (or of the needs to separate data subsets, each corresponding to a single state
direction of the stress vector) result in large variations in the direc- of stress (e.g. Angelier and Manoussis, 1980). As mentioned above
tion of the shear vector, and the shear-slip angle widely varies with- such dataset separation has to take into account field observations
out real significance for the misfit of the fault slip. Angelier (1990) when available (e.g. Armijo et al., 1982). But even so, one generally
solved this problem with the “Direct Inversion method” (INVD). In- has to test different dataset separations, until finding fault sub-sets
stead of simply considering the angle (α) on the fault plane between each one corresponding to a single state of stress.
the computed shear stress and actual slip vector (S, τ) (Fig. 6), he pro- We test herein several realistic and unrealistic situations inspired
posed a new criterion that allows simultaneously (1) minimizing the from fault diagrams found in literature to determine the best method
shear-slip angle and (2) having relative magnitude of shear stress to use according to the various fault slip datasets. We want to show,
large enough to induce slip despite rock cohesion and friction. This on one hand, how unrealistic fault set geometries can provide appar-
criterion named “RUP” for UPsilon Ratio, is the modulus of a vector ently good paleostress models if performed without care or without
υ (Fig. 6) between the computed shear stress vector (τ) and a vector
parallel to the measured striation (S). Without introducing rupture/

Fig. 6. Different criteria used in paleostress computation: α, for the classical methods
(e.g. R4DT method, Angelier, 1984) and υ, for the INVD method (Angelier, 1990). The
classical methods minimize, through 3D or iterative processes, a function of the angle
(α) between the computed (theoretical) shear stress (τ) and the actual slip vector
(S: measured striation). The INVD (Direct Inversion) method directly resolves equa-
tions where υ is the modulus of a vector between a unit vector along shear stress Fig. 7. Possible origins of discrepancies between fault measurements in paleostress
shear (τ), and a scalar product of a unit slip vector along striae (S). computation.
150 J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162

the appropriate computation method, and on the other hand, how For our tests we chose fault datasets with simple geometries which
specific fault geometries influence the results for each paleostress allow the results of paleostress computation to be also intuitively
computation method. Notice that the presence of fault data that will checked. These results are also reported in Table 1.
deviate the paleostress model from the real state of stress can result
not only from the presence of faults formed in differently oriented 3.2.1. Simple antithetic fault set geometries
stress fields (Fig. 7C) but also from secondary tilting of fault planes Angelier (1990) points out that with the classical methods, like
(Fig. 7A), or from inaccurate measurements of structural data his R4DT method, the orientation of the maximum and minimum
(Fig. 7B). principal stress axes in the case of conjugate faults is indeterminate
We used the INVD method (Angelier, 1990), and one of the in- (Fig. 8). Actually, the shear-slip angle on fault planes does not vary
version methods that uses the classical shear-slip criteria (Fig. 6): for the various orientations of the σ1 axis within the dihedron
the R4DT method (Angelier, 1984). We followed for each inver- shown in Fig. 8. In contrast, the INVD method, that considers the rel-
sion methods the quality criteria and their acceptable values as ative shear stress amplitude, would provide a unique and best orien-
defined by Angelier (1990). With the R4DT method, an individual tation of the paleostress axes (black arrow in Fig. 8). He concluded
fault with a shear-slip angle (ANG) lower than 22.5° is considered that the INVD method must be used instead of classical method in
as fitting very well the computed average stress tensor, individual the case of fault slip datasets with geometries close to conjugate
values between 45°and 22.5° may be consistent or not with the faults, so generally speaking in the case on mixed neoformed and
average solution depending on the range of uncertainties, whereas inherited faulting.
upper values are considered as not significantly fitting the average We compare the results of the INVD method with those of the
solution. For the INVD inversion method, fault slips with RUP values R4DT method for various dihedral angles between antithetic faults.
lower than 50% are considered as fitting the average solution very well, When cohesive material fails in a brittle manner, the angle between
those with values between 75 and 50% can be accepted, whereas those the conjugate faults (2θ), depends on the rock material according
with upper values must be rejected because not compatible with the to the Mohr–Coulomb failure criterion and of the stress values
modeled paleostress. The choice of these values is justified by results of (e.g., Chemenda, 2007). In rock types where fault slips are generally
stress inversion using data from sites with monophase deformation. measured (limestones, sandstones, basalts …) this value is commonly

Table 1
Results of the stress inversions presented in Figs. 11, 12, 13 and 18. σ1, σ2, σ3: maximum, intermediary and minimum principal stress axes respectively. Trend (north to east) and
plunge in degrees of the stress axes. Φ = (σ2 − σ3) / (σ1 − σ3). Quality estimator for the classical R4DT method: ANG = average angle between computed shear stress and observed
slickenside lineation (degrees) (Angelier, 1984). Quality estimator for the INVD method: RUP (0 ≤ RUP ≤ 200) taking into account the relative magnitude of the shear stress on fault
planes (Angelier, 1990).

Figure Diagram σ1 σ2 σ3 ANG RUP PHI


or test
Method Trend Plunge Trend Plunge Trend Plunge
name

Fig. 11 Fault 1 R4DT 77 1 270 89 167 0 1 0.19


Fault 1 INVD 271 2 137 88 1 2 2 14 0.46
Fault 2 R4DT 80 0 212 89 350 1 1 0.40
Fault 2 INVD 269 3 116 87 359 1 1 14 0.52
Fault 3 R4DT 278 1 146 89 8 1 1 0.85
Fault 3 INVD 267 5 104 85 351 2 3 16 0.36
Fault 4 R4DT 281 5 111 85 11 1 1 0.96
Fault 4 INVD 264 7 98 82 354 2 4 21 0.61
Fig. 12 CONJ1 INVD 268 0 358 0 176 90 0 10 0.50
R4DT 269 11 359 2 92 79 0 0.04
CONJ2 INVD 270 13 179 2 81 77 1 2 0.48
R4DT 270 13 179 2 81 77 0 0.16
CONJ3 INVD 278 9 186 16 36 72 1 7 0.41
R4DT 270 17 179 1 86 72 0 0.43
Fig. 13 CONJ4 INVD 88 0 358 11 180 79 1 12 0.50
R4DT 263 15 357 11 121 71 0 0.78
CONJ5 INVD 251 9 344 14 130 74 3 9
R4DT 25 20 122 17 250 63 0 0.98
CONJ6 INVD 87 1 357 21 180 69 1 11 0.51
R4DT 79 4 348 20 182 70 0 0.93
CONJ7 INVD 268 13 3 20 148 67 3 5 0.60
R4DT 61 19 347 12 205 68 0 0.95
CONJ8 INVD 89 1 358 30 180 60 0 11 0.50
R4DT 88 2 357 30 181 60 0 0.65
CONJ9 INVD 263 10 359 31 156 57 1 3 0.47
R4DT 96 13 358 29 207 57 0 0.33
CONJ10 INVD 89 1 358 40 180 50 0 11 0.50
R4DT 89 1 358 40 180 50 0 0.42
CONJ11 INVD 263 10 359 30 156 58 2 6 0.32
R4DT 263 11 1 36 159 52 0 0.45
CONJ12 INVD 273 3 6 45 180 45 1 12 0.49
R4DT 260 9 357 38 159 51 0 0.09
CONJ13 INVD 261 0 358 43 162 46 2 5 0.35
R4DT 243 21 350 37 130 45 0 0.05
CONJ14 INVD 89 1 357 60 180 30 0 11 0.50
R4DT 89 1 357 60 180 30 0 0.52
CONJ15 INVD 260 7 358 50 164 39 2 6 0.31
R4DT 97 3 2 58 189 32 0 0.23
Fig. 18 IPFW1 R4DT 340 73 237 4 146 16 9 0.41
J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162 151

We found that with the INVD method, the orientation of the max-
imum principal stress axis is always consistent with the symmetry of
the tested faults (E–W compression) and the value of the shape factor
Φ is generally 0.5. However, in the case of a very large dihedral angle
(170° Fig. 9D) the ratio Φ is low (0.26) which means that the values of
the σ2 and σ3 stresses are close together, and their orientation is not
constrained, which explains their plunge of 30 to 50° in Fig. 9D.
With the classical method R4DT (based on the minimization of the
shear-slip angles), the orientation of the stress axes deviates by 10° at
β = 120° (Fig. 9C) and the value of the shape factor Φ varies from
0.001 (Fig. 9A) to 0.99 (Fig. 9D). Notice that with this simple fault
geometry, the Φ ratio is not well constrained in any method. Whereas
remaining close to 0.5 with the INVD method, its value varies from
0.001 (for β = 10°) to 0.99 (for β = 170°) with the R4DT method
showing major consequences on the apparent fault compatibility
Fig. 8. Theoretical indeterminacy of the orientation of the σ1 axis with the classical and on the result (cf. below).
methods based on the minimization of the shear-slip angles (such as R4DT) compared For β close of 170° (Fig. 9D), the orientation of σ3 axes is con-
with the INVD method that uses the υ criterion (Fig. 6). Large arrows show the orien- strained for the two methods by the small dihedron between the
tation of the σ1 stress axes; in gray for the classical methods (R4DT…); in black for the fault planes, and the σ1 stress axis is therefore correctly oriented.
INVD method. The double-arrows curved line indicates the domain where the location
of the σ1 axis may vary with the R4DT method; the five-branch star shows the location
Using the R4DT method this result seems acceptable. With the INVD
of the σ1 axis with INVD method. method however, an anomalously large value of RUP (85%) indicates
that it corresponds to faults with low friction. The INVD method
allows discriminating this situation that would be unrealistic except
around 60°. In a rock mass where fractures pre-exist, the angle between for near frictionless faulting (e.g., weak faults, Zoback et al., 1987).
antithetic faults will also strongly depend on the friction of the fault For β close to 10° (Fig. 9A), the orientation of σ1 axis is constrained
planes. Without knowing if the faults are neoformed or inherited, we for the two methods by the small dihedron between the fault planes.
will call “β” the dihedral angle between the antithetic faults in the Here also the INVD method indicates an anomalously large value of
shortening dihedron. We tested the differences between the INVD RUP (83%) and the result can be rejected because the faults are close
inversion method and one of the classical inversion methods (R4DT) to tension fracture with no slip movement. The R4DT gives similar
for different values of the β angle (Fig. 9). For mathematical reasons, stress orientations, but with acceptable values of the ANG quality
in order to determine the four unknowns of the reduced stress tensor, factor. Such fault geometries with a small dihedron around σ1 are
software need at least four fault slip data. To allow the results to be common in natural examples because block rotations and stress
intuitively checked, we chose four fault planes, symmetrical with the deviations result in variations in fault planes attitudes (Fig. 7A). For
N–S and E–W axes, and corresponding to a strike-slip deformation example, concerning the fault data of site 5 (Fig. 2) with a small
with E–W shortening (Fig. 9). In practice, the tested geometries can value of the β angle, we choose the R4DT result for which the value
be natural, or can result from the artificial mixture of fault data from of the quality indicator is good for all the faults, because in the inver-
different state of stress (Fig. 7). The test has been carried out for β angles sion we wanted to use all the faults, even if some may have under-
of 10°, 60°, 120° and 170° (Fig. 9). gone small rotations explaining this unusual angle.

Fig. 9. Comparison between the R4DT and the INVD methods for simple antithetic fault set geometries and different values of the dihedral angle (β). Five-branch star = σ1 (max-
imum principal stress axis); four-branch star = σ2 (intermediate principal stress axis); three-branch star = σ3 (minimum principal stress axis). Empty stars: stress axes determined
with the INVD method. Infilled stars: stress axes determined with the R4DT method. Large arrows show the orientation of the σ1 and σ3 stress axes; in gray for the classical methods
(R4DT…); in black for the INVD method. ANG and RUP are the quality factors used for the R4DT and the INVD methods respectively (see Section 3.1).
152 J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162

Fig. 10. Example of paleostress inversion using the INVD method in the case of antithetic faults with a large dihedral angle. The site Cengle is located in the uppermost limestone of
the Late Cretaceous–Eocene Arc piggyback basin at the front of the Ste Victoire Mountain, near Aix-en-Provence, France (coordinates: 43°30.68′ N 5°36.75′ E).

These comparisons of various strike-slip geometries confirm that paleostress as corresponding to the sediment age, here a N–S com-
the σ1 stress axis direction is generally better constrained with the pression during Eocene.
INVD method than with the R4DT method. However, in this simple
symmetric example the R4DT solution has deviated from the INVD 3.2.2. Influence of a single oblique fault
solution of only 10°, in contrast to what could be expected (Fig. 8) As we found that the best fits between the two methods occur
and both methods provide similar stress directions. However, for a with the common β angle around 60° (Fig. 9B), we chose this angle
large β angle, and when near frictionless faulting can be considered to test the influence of an additional fault with different dips and
as realistic, the R4DT method might provide a better attitude of the strike that constrain the Φ ratio. We performed several tests with
σ2 and σ3 axes (Fig. 9D). strike-slip faults dipping 60° to the east, but with various strikes
A natural example of a large dihedral angle, where we choose the N50° (case 1), N40° (case 2), N30° (case 3), and N20° (case 4). Fig. 11
INVD method, is shown in Fig. 10 with the fracturing of a lacustrine illustrates the computed stress axes for these four cases, and provides
limestone on top of the Late Cretaceous–Eocene Arc basin near Aix- numerical results for the two extreme cases, 1 and 4 (the other re-
en-Provence. The Arc basin (Fig. 10) is a piggyback basin, and com- sults are reported in Table 1). In natural fault data similar 60° dipping
pressional fracturing occurred during sedimentation. In its upper- faults are frequently previous normal faults reactivated as strike-slip,
most rock formation all the measured faults show irregular surfaces or result from variations in the measurement of a non-planar sub-
suggesting unconsolidated sediments. In this case, the large dihedral vertical fault surface (Fig. 7B).
angle could result from fracturing at low confining pressure. Such Amongst the tested cases with different additional fault orien-
fault geometry can be used indirectly to determine the age of the tations (cases 1–4), the orientation of the σ1 axis varies by 24° with

Fig. 11. Influence of a single oblique fault on the stress inversion of conjugate faults for the R4DT (A) and the INVD (B) methods. For each test we used only one oblique fault. The
tested fault and the corresponding stress axes are numbered from 1 to 4. Results are reported in Table 1.
J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162 153

Fig. 12. Tests showing the differences in solution of inversion between fault datasets with antithetic faults (diagram CONJ1), without antithetic faults (diagram CONJ2), and with
irregular fault surfaces (diagram CONJ3).

the R4DT method (Fig. 11A), and only by 7° with the INVD method by the friction on the fault planes. In the case of reverse faults (Fig. 12)
(Fig. 11B). Compared with the stress computation with four strike-slip variations of this poorly constrained reduced stress tensor occur in the
faults of Fig. 9B, where compression is tending E–W, σ1 deviates 13° plunge of the σ1 and σ3 axes, and the Φ ratio goes close to zero.
(N77E) with the R4DT method but it only deviates 6° (N84E) with the With only one fault family (east dipping), the solution with the
INVD method (Fig. 11). We interpret the larger variations with the INVD method changes also because the plunge of the σ1 axis is only
R4DT method by the fact that with this method, the orientation of constrained by the friction on the east-dipping faults (Fig. 12 diagram
the stress axes is not constrained by the other faults (cf. supra). Con- CONJ2). In the tested fault set the INVD and R4DT solutions are very
versely, with the INVD method, the stress axes were constrained by similar.
the friction of the vertical faults and one additional fault has little As these results without antithetic fault are not well constrained,
influence on their orientation. The major deviation (13°) compared we tested the influence of a single additional oblique fault. We chose
with the result of Fig. 9B, is for an oblique fault oriented N50 (case 1), a reverse fault with an identical dip angle and the same azimuth of
and with the R4DT method (Fig. 11). striation, but with a different strike. Such oblique faults are com-
Notice that whereas in the INVD method the Φ ratio is relatively mon when measuring fault surfaces in a thrust zone with faults
stable, varying from 0.36 (fault N30) to 0.61 (fault N20), with the R4DT that might have rotated or fault surfaces that are listric or irregular
method this ratio increases from 0.19 (fault N50) to 0.96 (fault N20) at the scale of the compass. Such oblique fault measurement can
(Table 1). The extreme value of Φ (0.96) for a largely oblique (N20°) also result from uncertainties while measuring the strike of a low
trending fault (case 4) means that the magnitudes of σ2 and σ1 were dipping fault with a compass. In diagram CONJ3 (Fig. 12), the strike
close. Therefore, the orientation of the σ3 axis is mainly constrained by of the oblique fault is N30°E. This test is different from those of
the slip on the N20 trending fault, which results in the rotation of this Fig. 11 because the oblique fault is within the small fault dihedron,
axis toward this striation, and consequently a rotation also of the σ1 axis. and we have no antithetic fault. The stress computation reveals
This test demonstrates the stronger influence of a single oblique that the INVD method is more sensitive to this variation than the
fault when using the R4DT method. We conclude that when fault R4DT one (Fig. 12 diagram CONJ3). With this method, the trend
slip data are not numerous the use of the INVD method should be of σ1 has rotated clockwise of 8°, whereas with the R4DT method
preferred. the plunge of σ1 increases by 4°, but its trend has not changed.
We conclude that, if the attitude of the fault planes in the small
4. Pitfalls in paleostress reconstructions fault dihedron (around σ1) is not well constrained, the use of R4DT
might be preferred.
4.1. Importance of antithetic faults To show the necessity of measuring antithetic faults we compare the
results of the two methods of stress computations with, and without
Even when quality factors (ANG and RUP in Angelier's methods) antithetic faults, for fault set of various dip angles (Fig. 13). This test
suggest acceptable results of paleostress computations, the stress could correspond to reverse fault families that have been tilted by
axes can be not well constrained. This is the case when the fault successive folding, but that are either partially measured (no anti-
data do not include antithetic faults. These faults with opposite senses thetic fault) or completely measured. The horizontal at the time of
of movement are necessary to constrain the orientation of the com- faulting would be the dashed lines in the diagrams of Fig. 13. For
puted stress axes. Nevertheless, computations without antithetic each example, the value of northward tilt is indicated by the dip of
faults are common in publications. We tested how such lack of anti- this plane in dashed lines (10° to 60°). Notice that as we consider a
thetic faults data can change the results for the two computation very simple fault population, we have no oblique faults that con-
methods. We consider a simple case of reverse faults with north– strain the Φ ratio.
south strike, dipping east and west, and with an east–west azimuth With antithetic faults (Fig. 13 diagrams of the left side) and with
of striation (Fig. 12). As the computation needs at least four data we the INVD method the stress axes computed do not change significant-
took faults with four different dip values (25°, 30°, 35°, 40°). ly (maximum 3° around the E–W compression) for the tested differ-
Stress inversions for this simple symmetric configuration show ent values of tilting. With the R4DT method however, deviations
the σ1 axis horizontal with the INVD method, but plunging of 11° occur up to 11° from this E–W compression (Fig. 13 diagrams of the
with the R4DT method (Fig. 12 diagram CONJ1). The stress axes are left side). It is noticeable that these deviations with the R4DT method
therefore better constrained with the INVD method, in agreement occur for the examples CONJ4, CONJ6 and CONJ12 where the Φ ratio is
with Angelier's statements (Fig. 8), because it takes into account the either very low (0.09, σ2 close to σ3) or very high (0.93, σ2 close to
relative friction on each fault. As noticed above with strike-slip faults σ1). These examples show that the INVD method provides more sta-
(Fig. 9C), using the R4DT method the σ1 axis deviates from the sym- ble Φ ratios and more stable orientations of the stress axes than the
metrical axes because the reduced stress tensor is not constrained R4DT method.
154 J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162
J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162 155

Fig. 14. Comparison of fault diagrams without and with indication of the bedding planes. If bedding plane is not measured and not represented in the fault diagram, the interpre-
tation of the stress regime could be erroneous.

Without antithetic faults (Fig. 13 diagrams of the right side), the be interpreted either as extensional or as strike-slip, depending on
difference between the two methods is even more significant. In the the attitude of the bedding plane. Interpretation of paleostress com-
example CONJ5, whereas with the INVD method the σ1 axis deviates putation is therefore difficult and might even be impossible in rocks
of up to 17° from the solution with conjugates (19° from the expected whose eventual rotation cannot be determined, like igneous rocks.
E–W compression), with the R4DT method the trend of σ1 deviates of We conclude that when rock formations have been tilted, fault dia-
up to 58° (65° from the expected E–W compression). Here again the grams without representation (or at least indication) of the bedding
largest variations occur because the Φ ratio reaches extreme values planes should be considered with care.
with the R4DT method (Φ = 0.98 for CONJ5 and Φ = 0.05 for Ambiguities in fault interpretation are even more important when
CONJ13, Fig. 13). rocks have been rotated to 90°. Sub-vertical bedding planes are com-
This result again points out the importance of measuring antithet- mon in mountain belts in particular due to folding (e.g., Bergerat
ic faults in particular when using a classical method when the Φ ratio et al., 2010). With the exception of syn-depositional faults, in sub-
is not constrained by oblique faults. Even if the INVD method is more vertical sequences it might be impossible to determine if a faulting
stable for this feature, in publications, paleostress tensors computed event occurred before or after the tilt of the bedding plane. Fig. 15
without antithetic faults should be regarded as less reliable owing illustrates different types of fractures and state of stress that can be
to the large degree of uncertainties in the solution, even for a low found in vertical bedding plane and their two possible interpreta-
tilt (Fig. 13 site CONJ5). tions, post- or pre-tilt. For example, reverse faults in a vertical se-
quence (Fig. 15 line 3) can also be interpreted as pre-tilt extension,
4.2. Importance of bedding plane attitude even if, in this case a compression if more probable according to the
folding that rotated the rocks (Fig. 15). We conclude that there is
In the diagram CONJ15 (Fig. 13), the faults look like strike-slip rarely a unique interpretation of faulting in a vertical stratigraphic
faults, but in fact we tested tilted reverse faults, which cannot be sus- sequence and that fault analysis should be avoided in this situation.
pected if the horizontal at the time of faulting, is not indicated like in
the diagram CONJ14. The horizontal at the time of faulting is fre- 4.3. Bedding plane faults and wedge faults
quently the bedding plane. Of course this may not be the case and
faulting may occur in tilted rocks. But we observed that faulting Two faults of different attitude that intersect but slipped in the
typically occurred before folding or after folding, probably because same direction are commonly called wedge faults or drawer faults.
in between shortening was mainly accommodated by folding. Except In such features, the slip on one fault plane was guided by the motion
for vertical bedding where interpretation of faulting might be am- on the other and striation is parallel to the intersection of the fault
biguous, the attitude of bedding planes can provide important in- planes. Such structures are rare, but can be found in a fault zone
formation on the exact nature of the faults. where slips occurred on parallel fractures and when the fault slip con-
We want to emphasize the importance of displaying the bedding nects from one surface to another through pre-existing secant frac-
plane in the fault diagram. Fig. 14 illustrates the ambivalence of tures like the normal faults of Fig. 16A. Such guided fault striation, if
fault geometries if the bedding plane (here representing horizontal recognized should not be used in paleostress reconstructions because
at the time of faulting), dipping at 45° W or E, is not represented. In the slips do not obey the state of stress solely. The basic assumption of
the case of flat and steeply dipping antithetic faults (Fig. 14A), the the paleostress inversion methods that the slip on each fault plane
computed stress axes can be interpreted either as extensional or as has the direction and the sense of the shear stress on this plane
compressional, depending on the attitude of the bedding plane. In does not apply in such cases.
the case of steeply dipping antithetic faults (Fig. 14B), with the However, some drawer faults might not be recognized in the field
same ambiguity on the bedding plane, the computed stress axes can and introduced in the fault data accidentally. We tested the influence

Fig. 13. Differences between stress inversion of tilted reverse faults with or without antithetic faults (tilt varying from 10° to 60°). The horizontal at the time of faulting is indicated
as dashed lines. Large black arrows indicate the trend of σ1 for the INVD method. Large gray arrows indicate the trend of σ1 for the R4DT method. On the right, the stress inversion
has been performed without the west-dipping faults.
156 J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162

Fig. 15. Ambiguities in fault interpretation in a sub-vertical stratigraphic sequence. Reliable interpretation of faulting in a vertical sequence might be impossible.

of such a fault in the stress computation for the two methods. We in- because they do not satisfy to the basic assumption of the paleostress
troduced a low dipping drawer fault, with a striation parallel to the methods.
dextral fault, into the conjugate fault set of Fig. 16. With the R4DT
method and a 30° dipping fault (Fig. 16B) the computed stress axes 4.4. Stress permutations
deviate of 6°. Deviation is even more significant with a drawer fault
dipping of only 10° (almost perpendicular to the main strike-slip The term “stress permutation” was introduced to describe rela-
faults) (Fig. 16C). Stress inversion with the INVD methods appears tionship between state of stress with the same stress axes orientation
to be more stable and the σ1 axis remains close to E–W. However, but where the intermediate principal stress axis σ2 is replaced either
the value of the RUP quality criteria is high for the drawer fault and by the maximum compressional stress σ1 (permutation σ1/σ2), or by
these fault slips that do not conform to the stress-shear relationship the minimum stress σ3 (permutation σ2/σ3) (Angelier and Bergerat,
(Bott, 1959; Wallace, 1951) can be identified with the INVD method 1983; Hippolyte et al., 1992; Larroque et al., 1987). Generally, the
and withdrawn from the dataset. fault separation that leads to computation of state of stress related
These tests suggest favoring the use of the INVD method in the through stress permutation is justified by fault slip chronologies
case of suspected drawer faults. Similarly, the striations on bedding on fault planes. If stress inversion was realized with fault slip data
planes, that can be guided by strike-slip faults, and more generally corresponding to stress permutations, the stress tensors would
that result from folding, should not be used in paleostress inversion be characterized by extreme values of its shape ratio (Φ close
J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162 157

Fig. 16. Example of drawer (or transfer) fault in a normal fault zone (northern Turkey) (A) and tests of stress inversion in the case of strike-slip faults with a drawer fault dipping 30°
(B) and 10° (C). The drawer fault tested could be for example a bedding plane fault. Large black arrows for the INVD method. Large gray arrows for the R4DT method.

to 0 in case of permutation σ2/σ3, or close to 1 for permutation in lateral confining pressure) which may result from rock deforma-
σ1/σ2), and erroneous interpretations can be given like multidir- tion or from plate geodynamics (Bergerat et al., 2000; Hippolyte et al.,
ectional extension. 1992) and they are favored by contrasts and anisotropy in rock prop-
Interpretation of state of stress related through stress permutation is erties (Hu and Angelier, 2004).
not straightforward. Two states of stress can correspond to two succes-
sive tectonic phases, like the strike-slip state of stress (with σ1 trending 5. Particular applications
N–S and σ3 trending E–W) related to the Pyrenean–Provençal compres-
sion, followed by the Oligocene extension (σ2 trending N–S and σ3 5.1. Determining a paleo-verticality from paleostress analysis
trending E–W) that affected the West European platform (permutation
σ1/σ2, Bergerat, 1985, 1987). There is no unequivocal criterion to decide if a paleostress recon-
Two states of stress related through stress permutation can also struction is realistic or results from either mistakes in data separa-
occur during a same tectonic phase. Indeed, a fault study in middle- tion, or incorrect field observation, or approximation in computation
upper Quaternary sediments of southern Italy showed that stress method. But generally, if the faults are measured in geologically
permutations occurred in less than 460,000 years (nannoplankton young rocks, one may expect monophase deformation and the third
zones NN20-21; Hippolyte et al., 1992). The fault data separation basic assumption of stress inversion methods from fault-slip data
was supported by chronologies of striae. Three states of stress with satisfied (all fault slips correspond to a single common stress tensor).
same orientation of the horizontal maximum and minimum stress In the case of a monophase deformation, we commonly found values
axes are characterized. These homoaxial states of stress correspond of Φ that are not extreme, and one of the principal stress axes is
to σ1/σ2 stress permutations (extensional to strike-slip states of nearly vertical and the other two axes are nearly horizontal (if the
stress) and σ2/σ3 stress permutations (strike-slip to compressional bedding was not tilted; Hippolyte et al., 1992, 1994b). This stress
states of stress). Stress permutation was even “total” because an axes attitude results from gravity and the Earth's free surface,
extensional regime was followed by a compressional one with the which constrains one axis to be vertical, being dependent of rock
same orientation of the maximum and minimum horizontal stress weight and fluid pressure. Exceptions are rare and correspond to
axes. Such compression/tension changes have also been described particular situations, such as the vicinity of diapirs or large volcanic
within earthquake aftershock sequences (e.g. Garcia et al., 2002; centers or in accretionary prisms where large-scale thrusting occurs
Mercier and Carey-Gailhardis, 1989). In southern Italy, the stress (Angelier, 1994). It follows that if the stress solution does not con-
permutations are found along the ~ 20 km long Mt. Pollino fault form to this situation, one must question if the paleostress result is
that we interpret as a normal oblique fault bordering the N–S Crati realistic or not, and why.
graben to the north. These stress permutation are spatial (from one As one stress axis should be parallel to the gravitational force that
site to another) and temporal (changes trough time). In the studied acts close to vertical on the Earth surface, paleostress reconstruc-
case, they are related to the accommodation of the deformation tions can be used to determine the paleo-vertical and paleo-horizontal
along an irregular normal oblique fault. But more generally, such stress attitudes in a folded rock. One paleostress axis can be considered as
permutation may reflect changes in confining conditions (variations a record of the paleo-vertical and can be used as evidence of block
158 J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162

tilting following a deformation event. This record of paleo-vertical obtained diagram shows antithetic strike-slip faults, which is com-
can be used to determine the chronology of faulting events in folded mon fault geometry, with a NNE trend of compression (Fig. 17B).
rocks where pre-tilt paleostresses can be distinguished from post- The measured foreset beds dip now to the NW (Fig. 17B), which is
tilt paleostresses (e.g. Bergerat et al., 2007, 2010; Hippolyte and consistent with a progradation of the infill of the canyon to the
Sandulescu, 1996; Vandycke and Bergerat, 2001; Yamaji et al., 2005). West as proposed by Clauzon et al. (1996). We conclude that the
This record of paleo-vertical is also useful in non bedded rocks like Messinian paleodrainage network in this area was characterized by
conglomerates or granites, where the attitude of the stress axes a confluence of the Paleo-Vésubie with a river flowing to the West
can reveal tectonic tilts (Hippolyte et al., 2011). We provide here from Tourrette, and by a confluence with the Paleo-Var River around
an example of such use in foreset conglomerates at the front of the Castagniers (Fig. 17B). Our paleostress analysis allowed character-
Southwestern Alps near Nice (Fig. 17). izing the attitude of a paleo-horizontal because strike-slip faulting
In foreset beds, the dip-direction of the bedding planes generally predates the tectonic tilt of the foreset beds on the St. Blaise Aspre-
indicates the direction of talus progradation. In the Nice area Clauzon mont thrust ramp.
et al. (1996) and Fauquette et al. (1999) reconstructed the Messinian This example illustrates that the use of paleostresses as markers
paleodrainage network of the Var and Vésubie rivers (Fig. 17). During of the paleo-horizontal is of particular interest in structural analy-
the Messinian desiccation event of the Mediterranean Sea, deep sis. It can be used in non horizontal sedimentation like the foreset
canyons were incised by rivers. After the Zanclean flooding of the beds, to reveal tectonic deformation or reconstruct directions of
Mediterranean, these canyons were filled, by prograding Gilbert progradation. Note that to reconstruct the paleo-horizontal we
type deltas characterized by marine foreset sediments. The Messinian– used tectonic faults of an early tectonic event, but compaction nor-
Pliocene confluence of the Vésubie and Var rivers was supposed to be mal faults, that can be found in clays or in conglomerate, can also
located around Castagnier and Tourrette villages (Clauzon et al., 1996) be used (Hippolyte et al., 2011). Such faults are of particular inter-
(Fig. 17). However, this area is also the western front of the Nice thrust, est because they form early in the rock and tectonic history and
represented by the Saint Blaise-Aspremont fault (Fig. 17), and the foreset clearly result from vertical gravitational forces.
beds have probably been tilted by its Neogene tectonic deformation As mentioned above, for this application of the paleostress inver-
(Larroque et al., 2011). East of the St. Blaise-Aspremont fault, the fore- sion, one must be careful if the value of Φ is close to 1 or to 0. For ex-
sets of conglomerate dip to the East of up to 48°, which could suggest ample in sites 13 and 22 (Fig. 2) where the stress inversion gives a
that the progradation of the Pliocene Vésubie was toward the east low value of the Φ ratio (cf. small size of the three-branch and four-
and that the confluence with the Pliocene Var was not near Castag- branch stars), the values σ2 and σ3 are close together and the plunge
niers (Fig. 17). of these stress axes is not constrained and therefore these stress axes
We measured fault planes at this key area to determine if tec- cannot be resolved to a paleo-vertical or a paleo-horizontal.
tonic tilt has occurred since the deposit of the Pliocene foreset
beds. We found reverse and vertical faults, which are not typical
geometries and that may have been tilted (Fig. 17A). The attitude 5.2. Characterization of structure
of the foreset beds cannot be measured accurately in conglomer-
ates, but they dip mainly to the east (Fig. 17A). Paleostress com- Mesoscale fault analysis and paleostress computation are particu-
putation indicates a NNE–SSW compression consistent with a larly useful in structural analysis. Many basins or mountain belts have
reverse dextral movement on the St. Blaise-Aspremont fault (Ritz, a long history with polyphase deformation, and the observed final
1991) (Fig. 17A). Using the paleostress axes we could determine a structure results from successive events that can be unraveled by
possible paleo-horizontal plane including the σ1 and σ3 stress axes analyzing mesoscale fault datasets. Even for simple structure paleos-
(Fig. 17A). We rotated the faults and bedding planes around the tress analysis can prove to be useful. For example, when the slip of
trend of the supposed paleo-horizontal by 73° to the WNW to restore a large fault revealed by mapping cannot be determined by direct
this paleo-horizontal to its initial position (during faulting). The field observation, because the soft material of the fault zone is often

Fig. 17. Use of stress inversion method for determining the paleo-vertical and the direction of talus progradation in tilted foreset beds infilling Messinian Canyon, Nice area, south-
ern Western Alps, France (coordinates: 43°48′ N 7°15′ E). Geological sketch of the Pliocene foreset sediments. Block diagrams and fault diagrams showing the relationship between
the stress axes and the bedding planes. A: Present attitude. Fault site West of Tourrette with σ2 as paleo-vertical, and paleo-horizontal including the σ3 and σ1 stress axes. B: After
back tilting, the dip of the foreset beds to the West indicates the direction of talus progradation in the Messinian canyons of Tourrette during the Pliocene. S0: measured foreset beds.
H: paleo-horizontal.
J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162 159

covered by vegetation, it can be inferred from paleostress analysis of 5.3. Age and chronology of structures
mesoscale faults found along the fault zone (e.g. Fig. 2).
Even when the striation can be observed on a large fault, paleos- Paleostress analysis using mesoscale faults can also be used in
tress analysis can be useful to determine the exact nature of this structural geology to date, or constrain the chronology, of larger
fault. Fig. 18 illustrates the case of a steeply dipping reverse fault in structures, or to elucidate the formation of polyphase structures
the southern Apennines. The Mesozoic limestone is found above (e.g. Hippolyte et al., 1994a, 1994b). Large-scale structures can be
Tortonian sandstones. However, the stratigraphic layers are tilted dated by correlation with a stratigraphically dated state of stress
and this apparent reverse fault could either be a reverse fault formed (Angelier, 1994; Hippolyte et al., 1993) or can be directly dated if
during folding, or a normal fault that has been tilted and that should syn-depositional faulting is found. In extensional deformation syn-
be considered as an extensional fault. The difference between these sedimentary normal faults are common (Fig. 4) (e.g. Hippolyte et
two interpretations is based on the type of secondary faults associated al., 1991). In compressional deformation one can use mesoscale faults
with this large fault as illustrated in the block diagrams (Fig. 18). If the present in growth strata to characterize syn-depositional paleos-
paleostress computation, from the main fault and the secondary faults, tresses and date the corresponding large scale structure.
indicates a compression (with σ1 horizontal), this fault is a reverse- Measuring fault slip in growth strata is of great interest because it
compressional fault. If it indicates a σ1 axis perpendicular to the can provide a direct dating of the paleostresses. For this purpose we
bedding plane, like in the fault diagram of Fig. 18, one may conclude need some fault slips predating, and others postdating, the synsedi-
that it is an extensional fault, even if it has now been tilted to a mentary tilt. This approach using the recorded paleo-horizontal, was
reverse fault geometry. Notice that this extensional fault might illustrated in late-Quaternary growth strata along a strike-slip fault
have been tilted by listric normal faults (Fig. 7A) or during folding. in Southern Italy (Hippolyte et al., 1992). Here we illustrate this
Here again mesoscale faults and paleostress analysis can allow the method in a compressional setting, at the southern front of the Sainte-
interpretation of the final structure. Victoire Mountain, where it also supports the chronology of large
scale structures (Fig. 19).
The Sainte-Victoire Mountain is located near Aix-en-Provence
at the northern front of the Pyrenean–Provençal Late Cretaceous–
Eocene fold-and-thrust belt. This E–W structure constitutes also
the northern border of the Arc piggyback basin (Fig. 10). The defor-
mation of the border of this piggyback basin is attested by outstand-
ing growth strata in particular in the Campanian–Maastrichtian
continental breccias, famous for dinosaur eggs, and in the Early
Danian continental breccias (Fig. 19; Leleu et al., 2005; 2009).
We measured faults in the Campanian breccias that dip 72° to the
north (Fig. 19). We found normal and strike-slip faults (Fig. 19, site
MARBW). The normal faults have striations parallel to the bedding
planes and might represent strike-slip faults that have been tilted.
The restoration of the bedding plane to the horizontal confirms that
these faults are strike-slip faults that have been tilted (Fig. 19, site
MARBW, diagram in the middle). They allow computation of a SSE-
trending compression.
The strike-slip faults measured in this site are sub-vertical with
horizontal striations that have apparently not been tilted (Fig. 19,
site MARBW, diagram on the right). They also allow computation of
a SSE-trending compression. Considering that the state of stress was
a SSE-trending compression before the tilting and after the tilting,
and that this compression is perpendicular to the trend of the bed-
ding, it is reasonable to conclude that this compression is responsible
for the folding. At this place the growth strata show bedding planes
varying progressively from sub-vertical to sub-horizontal, suggest-
ing that the folding occurred during the Campanian–Maastrichtian.
It is therefore possible to date the SSE-trending compression of the
Campanian–Maastrichtian, and to conclude that the Sainte-Victoire
Mountain is a compressional structure active at least in the Late
Cretaceous under this compression.
At the same site, mesoscale fault analysis can be used to support
the chronology of other large structures. The Sainte-Victoire Moun-
tain and the Arc basins are cut to the west by NNE-trending normal
faults including the Aix-en-Provence fault and the Meyreuil fault
(Fig. 10). The Aix-en-Provence fault mainly moved during the Oligo-
cene and border the Aix-en-Provence half graben that is part of the
West European rift system (Hippolyte et al., 1993). In the Sainte-
Victoire Mountain we also found the Roques Hautes fault that cuts
the Late Cretaceous growth strata (Figs. 10 and 19). This fault could
be either a transfer fault, active during compression, or an Oligocene
Fig. 18. Example of use of mesoscale fault analysis for characterizing a larger scale normal fault belonging to the NNE-trending normal fault system of
structure, Matese Mountains, Southern Apennines, Italy (coordinates: 41°24.9′ N
14°31.5′ E) (modified from Hippolyte et al., 1995). Whether this reverse fault is a com-
the Aix-en-Provence Oligocene half graben that develops to the West.
pressional or an extensional fault that has been tilted, can be determined by analyzing Measurement of fault slips along the Roques Hautes fault and in
the associated mesoscale faults and corresponding paleostress axes. the sub-horizontal Maastrichtian formation reveals N–S trending
160 J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162

Fig. 19. Dating paleostresses using pre-tilt and post-tilt faults in growth stratas, and chronology of large scale structures (Late Cretaceous–Eocene Pyrenean–Provençal fold and
thrust belt, Oligocene Aix-en-Provence basin) from polyphase faulting in syn-tectonic deposits and mesoscale fault chronology; front of the Ste Victoire Mountain, location in
Fig. 10 (coordinates: 43°31.6′ N 5°32.8′ E). Site MARBW in the tilted Campanian formation is cut by pre-tilt and post-tilt strike-slip faults corresponding to NNW–SSE compression
contemporaneous of the tilting of the Maastrichtian growth strata. Site Roques1 shows normal faults of ESE–WNW extension postdating the Maastrichtian syn-compressional
growth strata. Site Arc3 shows polyphase deformation with ESE–WNW extension post-dating N–S compression as attested by chronology of slickenside superposition (reverse
and normal slip on a south dipping fault surface).

faults corresponding to an E–W extension (Fig. 19, site Roques1). To sites and by structural analysis that will be the object of a more de-
the south another site of fault measurements confirms that an tailed publication.
extensional state-of stress affected this area (Fig. 19, site Arc3). In
this later site, a previous E–W reverse fault was reactivated as an
oblique normal fault during the E–W extension as shown by the 6. Conclusions
superposed striations. We conclude that E–W extension post-
dated SSW-trending compression and that this chronology of faults Above all, the importance of accurate observations and measure-
confirms the chronology of the structures: (1) formation of the ments during the field data acquisition is underlined. Poor recogni-
Sainte-Victoire thrust system and Arc basin from Late Cretaceous to tion of the sense criteria, unawareness of bedding attitude, and lack
Eocene; (2) opening of the Aix-en-Provence basin during the Oligocene. of relative chronological observations may bring to false interpreta-
Notice that the normal faults and their chronology confirm the interpre- tions of the faulting episodes even if the stress computations are
tation of Lacombe et al. (1992) based on paleostress analysis from mathematically correct.
calcite twins, that the E–W Oligocene extension affected the Ste Victoire The most important results revealed by a series of comparative
Mountain. Of course, even if we choose these sites to illustrate the tests carried out with one of the classical inversion methods based
information that mesoscale faults can give on structures, our general on the minimization of the shear-slip angles (R4DT; Angelier, 1984)
conclusions about the Sainte-Victoire system and Arc basin are not and the “Direct Inversion” method (INVD; Angelier, 1990) to compute
only founded on these three sites but are supported by many other the reduced stress tensor, may be summarized as follows.
J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162 161

(1) The INVD method must be preferred to the classical R4DT and seismotectonics in the foreland of the Southern Apennines. Geological Society
of America Bulletin 121, 1421–1440.
method in case of simple antithetic faults (in particular be- Armijo, R., Cisternas, A., 1978. Un probleme inverse en microtectonique cassante. C. R.
cause it allows identification of abnormal values of friction) Acad. Sc., Paris 287, 595–598.
and in case of oblique faults. Armijo, R., Carey, E., Cisternas, A., 1982. The inverse problem in microtectonics and the
separation of tectonic phases. Tectonophysics 82, 145–160.
(2) Inconsistencies between these two methods are typically no- Bellier, O., Zoback, M., 1995. Recent state of stress change in the Walker Lane zone
ticed in the value of the shape factor (Φ). The value of Φ is western basin and Range Province-USA. Tectonics 14, 564–593.
highly variable with the R4DT method. It might cause difficul- Bergerat, F., 1985. Déformations cassantes et champs de contrainte tertiaires dans la
plate-forme européenne. Thèse Doct. Etat ès-Sciences, Univ. P.& M. Curie, Paris.
ties, because for values close to 0 or 1, the number of faults in Mém. Sc. Terre 85-07, 315 p.
a fault population with acceptable shear-stress angle generally Bergerat, F., 1987. Stress fields in the European Platform at the time of Africa–Eurasia
increases, and incompatible faults might appear compatible. collision. Tectonics 6, 99–132.
Bergerat, F., 1994. From inversion methods to paleostress fields reconstructions in plat-
(3) The results of stress inversion from fault datasets without anti-
forms, chains and basins: an overview. Some examples in Western and Central Europe.
thetic faults can be highly variable, in particular with the R4DT In: Roure, F. (Ed.), Peri-Tethyan Platforms. Technip, Paris, pp. 159–178.
method. However, if there are oblique faults, the R4DT method Bergerat, F., Angelier, J., Homberg, C., 2000. Tectonic analysis of the Husavik-Flatey
can provide reliable results. Fault (northern Iceland) and mechanisms of an oceanic transform zone, the
Tjörnes Fracture Zone. Tectonics 19, 1161–1177.
(4) Faults measured in tilted rocks should always be shown in Bergerat, F., Angelier, J., Andreasson, P.G., 2007. Heterogeneous Paleostress field and
Schmidt's diagrams with an accompanying bedding plane be- brittle deformation of the Tornquist Zone in Scania (Sweden) during Permo-
cause it may change the interpretation of paleostress result. Mesozoic and Cenozoic times. Tectonophysics 444, 93–110.
Bergerat, F., Vangelov, D., Dimov, D., 2010. Brittle deformation, paleostress field reconstruc-
(5) Interpretation of faults in a vertical sequence is hazardous. tion and tectonic evolution of the Eastern Balkanides (Bulgaria) during Mesozoic and
Cenozoic times. In: Sosson, M., Kaymakci, N., Stephenson, R.A., Bergerat, F., Starostenko,
Finally, when the fault datasets are of good quality, and when V. (Eds.), Sedimentary Basin Tectonics from the Black Sea and Caucasus to the Arabian
Platform. : Special Publication, 340. Geological Society, London, pp. 77–111.
the correct inversion method is chosen, fault analysis and paleos- Bott, M.H.P., 1959. The mechanisms of oblique slip faulting. Geological Magazine 96,
tress computation can be achieved in highly deformed areas such 109–117.
as mountain belts. Stress inversion appears to be a powerful tool Carey, E., 1976. Analyse numérique d'un modèle mécanique élémentaire appliqué à
l'étude d'une population de failles: calcul d'un tenseur moyen des contraintes à
in structural geology not only by characterizing the paleo-states of partir des stries de glissement. Unpublished PhD Thesis, Univ. Paris Sud, Orsay,
stress, but also by permitting interpretation (nature and sense of 138 p.
faults) and dating of large structures. When in an area, the stresses Carey, E., 1979. Recherche des directions principales de contrainte associées au jeu d'une
population de faille. Revue de Géologie dynamique et de Géographie physique 2,
changed through time and when these changes are dated using the
57–66.
stratigraphic sequence, characterizing a structure in terms of paleos- Carey, E., Brunier, B., 1974. Analyse théorique et numérique d'un modèle mécanique
tresses can allow its dating through correlation to the known stress élémentaire appliqué à l'étude d'une population de failles. C.R. Acad. Sci., Paris
evolution. Moreover, as shown by the typical coincidence of a 279, 891–894.
Carey-Gailhardis, E., Mercier, J.L., 1987. A numerical method for determining the state
stress axes with the vertical during faulting, we showed that the of stress using focal mechanisms of earthquake populations: application to Tibetan
computed paleostress axes can be used as a record of the paleo- teleseisms and microseismicity of Southern Peru, Earth Planet. Science Letters 82,
horizontal and paleo-vertical. In particular, their attitude can be used 165–179.
Célérier, B., 1995. Tectonic regime and slip orientation of reactivated faults. Geophysical
to determine the chronology of tectonic events, to reveal tilts in non Journal Intaernational 121, 143–161.
bedded rocks, and in tilted foreset beds, to infer a direction of sediment Chemenda, A.I., 2007. The formation of shear-band/fracture networks from a constitu-
progradation. tive instability: theory and numerical experiment. Journal of Geophysical Research
112, B11404. doi:10.1029/2007JB005026.
Clauzon, G., Rubino, J.-L., Suc, J.-P., 1996. Les rias Pliocènes du Var et de Ligurie: comble-
Acknowledgments ment sédimentaire et évolution géodynamique. Field Trip Guide Book of Groupe
Français d'Etude du Néogène/Groupe Français de Géomorphologie, p. 36 p.
Delvaux, D., Sperner, B., 2003. Stress tensor inversion from fault kinematic indicators
The authors are grateful to Dr. Emanuele Fontana and Dr. Steffen and focal mechanism data: the TENSOR program. In: Nieuwland, D. (Ed.), New In-
H. Büttner for helpful reviews. sights into Structural Interpretation and Modelling. : Special Publication, 212.
Geological Society, London, pp. 75–100.
Doglioni, C., Tropeano, M., Mongelli, F., Pieri, P., 1996. Middle-Late Pleistocene uplift of
References Puglia: an “anomaly” in the apenninic foreland. Memorie della Società Geoligica
Italiana 51, 101–117.
Angelier, J., 1975. Sur un apport de l'informatique à l'analyse structurale: exemple de la Etchecopar, A., Vasseur, G., Daignières, M., 1981. An inverse problem in microtectonics
tectonique cassante. Revue de Géologie dynamique et de Géographie physique 17, for the determination of stress tensor from fault striation analysis. Journal of
137–146. Structural Geology 3, 51–64.
Angelier, J., 1979. Determination of mean principal stresses for a given fault population. Fauquette, S., Clauzon, G., Suc, J.-P., Zheng, Z., 1999. A new approach for paleoaltitude
Tectonophysics 56, T17–T26. estimates based on pollen records: example of the Mercantour Massif (southeastern
Angelier, J., 1984. Tectonic analysis of fault slip data sets. Journal of Geophysical Research France) at the earliest Pliocene. Earth and Planetary Science Letters 170, 35–47.
89, 5835–5848. Funiciello, R., Montone, P., Salvini, F., Tozzi, M., 1988. Caratteri strutturali del promon-
Angelier, J., 1990. Inversion of field data in fault tectonics to obtain the regional stress. torio del Gargano. Memorie della Societa Geologica Italiana 41, 1235–1243.
III—a new rapid direct inversion method by analytical means. Geophysical Journal Garcia, S., Angelier, J., Bergerat, F., Homberg, C., 2002. Tectonic analysis of an oceanic
International 103, 363–376. transform fault zone revealed by fault-slip data and earthquake focal mechanisms:
Angelier, J., 1994. Fault slip analysis and palaeostress reconstruction. In: Hancock, P.L. the Husavik-Flatey Fault, Iceland. Tectonophysics 344, 157–174.
(Ed.), Continental Deformation. Pergamon Press, pp. 53–100. Gephart, J., Forsyth, D.W., 1984. An improved method for determining the regional stress
Angelier, J., 2002. Inversion of earthquake focal mechanisms to obtain the seismotec- tensor using earthquake focal mechanism data: application to the San Fernando
tonic stress IV—a new method free of choice among nodal planes. Geophysical earthquake sequence. Journal of Geophysical Research 89, 9305–9320.
Journal International 150, 588–609. Hardcastle, K.C., 1989. Possible paleostress tensor configurations derived from fault-slip
Angelier, J., Bergerat, F., 1983. Systèmes de contrainte en extension intracontinentale. data in eastern Vermont and western New Hampshire. Tectonics 8, 265–284.
Bulletin du Centre de Recherche Exploration et Production Elf-Aquitaine 7, Hardcastle, K.C., Hills, L.S., 1991. Brute3 and Select: Quickbasic 4 programs for determi-
137–147. nation of stress tensor configurations and separation of heterogeneous populations
Angelier, J., Goguel, J., 1979. Sur une méthode simple de détermination des axes princi- of fault-slip data. Computers and Geosciences 17, 23–43.
paux des contraintes pour une population de failles. C. R. Acad. Sc., Paris (D) 288, Hippolyte, J.-C., 1992. Tectonique de l'Apennin méridional : structures et paléocon-
307–310. traintes d'un prisme d'accrétion continental. Unpublished Ph.D. thesis, Univ. P.&
Angelier, J., Manoussis, S., 1980. Classification automatique et distinction de phases M. Curie, Paris. Mém. Sc. Terre 92–5, 318 p.
superposées en tectonique cassante. C. R. Acad. Sc., Paris 290, 651–654. Hippolyte, J.-C., Sandulescu, M., 1996. Paleostress characterization of the ‘Wallachian
Angelier, J., Mechler, P., 1977. Sur une méthode graphique de recherche des contraintes phase’ in its type area (southeastern Carpathians, Romania). Tectonophysics 263,
principales également utilisable en tectonique et en séismologie: la méthode des 235–248.
dièdres droits. Bulletin de la Société géologique de France 19, 1309–1318. Hippolyte, J.-C., Nury, D., Angelier, J., Bergerat, F., 1991. Relations tectonique extensive
Argnani, A., Rovere, M., Bonazzi, C., 2009. Tectonics of the Mattinata fault, offshore et sédimentation continentale: exemple des bassins oligocènes de Basse-Provence.
south Gargano (southern Adriatic Sea, Italy): implications for active deformation Bulletin de la Société géologique de France 162, 1083–1094.
162 J.-C. Hippolyte et al. / Tectonophysics 581 (2012) 144–162

Hippolyte, J.-C., Angelier, J., Roure, F., 1992. Les permutations de contraintes dans un orogène: of the 1978 Thessaloniki (Greece) and 1980 Campania-Lucania (Italia) earth-
exemple des terrains quaternaires du sud de l'Apennin. C. R. Acad. Sc., Paris 315, 89–95. quakes as example. Earth and Planetary Science Letters 92, 247–264.
Hippolyte, J.-C., Angelier, J., Nury, D., Bergerat, F., Guieu, G., 1993. Tectonic-stratigraphic Michael, A.J., 1984. Determination of stress from slip data: faults and folds. Journal of
record of paleostress time changes in the Oligocene basins of the Provence, southern Geophysical Research 89, 11517–11526.
France. Tectonophysics 226, 15–35. Oncken, O., 1988. Aspects of the reconstruction of the stress history of a fold and
Hippolyte, J.-C., Angelier, J., Roure, F., 1994a. A major geodynamic change revealed by Quater- thrust belt (Rhenish Massif, Federal Republic of Germany). Tectonophysics 152,
nary stress patterns in the Southern Apennines (Italy). Tectonophysics 230, 199–210. 19–40.
Hippolyte, J.-C., Angelier, J., Roure, F., Casero, P., 1994b. Piggyback basin development Orife, T., Lisle, R.J., 2006. Assessing the statistical significance of paleostress estimates:
and thrust belt evolution: structural and paleostress analyses of Plio-Quaternary simulations using random fault-slips. Journal of Structural Geology 28, 952–956.
basins in the Southern Apennines. Journal of Structural Geology 16, 159–173. Otsubo, M., Sato, K., Yamaji, A., 2006. Computerized identification of stress tensors de-
Hippolyte, J.-C., Angelier, J., Barrier, E., 1995. Compressional and extensional tectonics termined from heterogeneous fault-slip data by combining the multiple inverse
in an arc system: example of the southern Apennines. Journal of Structural geology method and k-means clustering. Journal of Structural Geology 28, 991–997.
17, 1725–1740. Petit, J.P., 1976. La zone de décrochement du Tizi N'Test (Maroc) et son fonctionnement
Hippolyte, J.-C., Mann, P., Grindlay, N., 2005. Geologic evidence for the prolongation of ac- depuis le Carbonifère. Unpublished PhD Thesis, Montpellier University, France, 99 p.
tive normal faults of the Mona Rift into northwestern Puerto Rico. In: Mann, P. (Ed.), Petit, J.P., 1987. Criteria for the sense of movement on fault surfaces in brittle rocks.
Active Tectonics and Seismic Hazards of Puerto Rico, the Virgin Islands, and Offshore Journal of Structural Geology 9, 597–608.
Areas. : Special Paper, 385. Geological Society of America, pp. 161–171. Petit, J.P., Proust, F., Tapponnier, P., 1983. Critères de sens de mouvement sur les miroirs de
Hippolyte, J.-C., Clauzon, G., Suc, J.-P., 2011. Messinian-Zanclean canyons in the Digne failles en roches non calcaires. Bulletin de la Société géologique de France 25,
nappe (southwestern Alps): tectonic implications. Bulletin de la Société géologique 589–608.
de France 182 (2), 111–132. Reches, Z., 1987. Determination of the tectonic stress tensor from slip along faults that
Hu, J.-C., Angelier, J., 2004. Stress permutations: three-dimensional distinct element obey the Coulomb yield condition. Tectonics 6, 849–861.
analysis account for a common phenomenon in brittle tectonics. Journal of Riedel, W., 1929. Zur Mechanik Geologischer Brucherscheinungen. Zentral-blatt fur
Geophysical Research 109, B09403. doi:10.1029/2003JB002616. Mineralogie, Geologie und Paleontologie B 354–368.
Lacombe, O., Angelier, J., Laurent, P., 1992. Determining paleostress orientations from Ritz, J.-F., 1991. Evolution du champ de contraintes dans les Alpes du Sud depuis la fin de
faults and calcite twins: a case study near the Sainte-Victoire Range (southern l'Oligocène. Implications sismotectoniques. Unpublished PhD Thesis, Montpellier II
France). Tectonophysics 201, 141–156. University, France, 187 p.
Larroque, J.M., Etchecopar, A., Philip, H., 1987. Evidence for the permutation of stress σ1 Schettino, A., Turco, E., 2011. Tectonic history of the western Tethys since the Late
and σ2 in the Alpine foreland: the example of the Rhine graben. Tectonophysics Triassic. Geological Society of America Bulletin 123, 89–105.
144, 315–322. Servizio geologico d'Italia, 1965, 1969, 1976. Carta geological d'Italia, 1/100.000. Servizio
Larroque, C., Delouis, B., Hippolyte, J.C., Deschamps, A., Lebourg, T., Courboulex, F., Bellier, geologico d'Italia. : Folio, 156. Organo cartographico dello stato, Roma, pp. 157–164.
O., 2011. Joint multidisciplinary study of the Saint-Sauveur-Donareo fault (lower Var Shabanian, E., Bellier, O., Abbassi, M.R., Siame, L., Farbod, Y., 2010. Plio-Quaternary
valley, French Riviera): a contribution to seismic hazard assessment in the urban area stress states in NE Iran: Kopeh Dagh and Allah Dagh-Binalud mountain ranges.
of Nice. Bulletin de la Société géologique de France 182 (4), 323–336. Tectonophysics 480, 280–304.
Leleu, S., Ghienne, J.F., Manatschal, G., 2005. Upper Cretaceous–Palaeocene basin-margin Shan, Y., Fry, N., 2005. A hierarchical cluster approach for forward separation of hetero-
alluvial fans documenting interaction between tectonic and environmental processes geneous fault/slip data into subsets. Journal of Structural Geology 27, 929–936.
(Provence, SE France). In: Harvey, A.M., Mather, A.E., Stokes, M. (Eds.), Alluvial Fans: Shan, Y., Fry, N., 2006. The moment method used to infer stress from fault/slip data in sigma
Geomorphology, Sedimentology, Dynamics. : Special Publication, 251. Geological space: invalidity and modification. Journal of Structural Geology 28, 1208–1213.
Society, London, pp. 217–239. Shan, Y., Lin, G., Li, Z., Zhao, C., 2006. Influence of measurement errors on stress estimated
Leleu, S., Ghienne, J.F., Manatschal, G., 2009. Alluvial fan development and morpho- from single-phase fault/slip data. Journal of Structural Geology 28, 943–951.
tectonic evolution in response to contractional fault reactivation (Late Cretaceous– Simòn-Gòmez, J.L., 1986. Analysis of a gradual change in stress regime (examples from
Palaeocene), Provence, France. Basin Research 21, 157–187. eastern Iberian Chain, Spain). Tectonophysics 124, 37–53.
Lisle, R., 1988. ROMSA: a basic program for paleostress analysis using fault striation Sperner, B., Zweigel, P., 2010. A plea for more caution in fault–slip analysis. Tectonophysics
data. Computers and Geosciences 14, 255–259. 482, 29–41.
Lisle, R., 1989. Paleostress analysis from sheared dike sets. Geological Society of America Twiss, R.J., Unruh, J.R., 1998. Analysis of fault slip inversions: do they constrain stress or
Bulletin 101, 968–972. strain rate? Journal of Geophysical Research 103, 12205–12222.
Lisle, R.J., Srivastava, D.C., 2004. Test of the frictional reactivation theory for faults and Vandycke, S., Bergerat, F., 2001. Brittle tectonic structures and palaeostress analysis
validity of fault-slip analysis. Geology 32 (7), 569–572. in the Isle of Wight, Wessex basin, Southern U.K. Journal of Structural Geology 23,
Lisle, R.J., Orife, T.O., Arlegui, L., Liasa, C., Srivastava, D.C., 1998. Favoured states of 393–406.
palaeostress in the Earth's crust: evidence from fault-slip data. Journal of Structural Vialon, P., Ruhland, M., Grolier, J., 1976. Elements de tectonique analytique. Masson,
Geology 28, 1051–1066. Paris. 118 pp.
Mann, P., Calais, E., Ruegg, J.C., DeMets, C., Jansma, P., Mattioli, G., 2002. Oblique collision in Wallace, R.E., 1951. Geometry of shearing stress and relation to faulting. Journal of Geology
the northeastern Caribbean from GPS measurements and geological observations. 59, 118–130.
Tectonics 21, 1057. doi:10.1029/2001TC001304. Will, T.M., Powell, R., 1991. A robust approach to the calculation of paleostress fields
Mann, P., Prentice, C., Hippolyte, J.-C., Grindlay, N., Abrams, L., Lao-Davila, D., 2005. Re- from fault plane data. Journal of Structural Geology 13, 813–821.
connaissance study of Late Quaternary faulting along Cerro Goden fault zone, Yamaji, A., Tomita, S., Otsubo, M., 2005. Bedding tilt test for paleostress analysis. Journal of
western Puerto Rico. In: Mann, P. (Ed.), Active Tectonics and Seismic Hazards of Structural Geology 27, 161–170.
Puerto Rico, the Virgin Islands, and Offshore Areas. : Special Paper, 385. Geological Zoback, M.D., Zoback, M.L., Mount, V.S., Suppe, J., Eaton, J.P., Healy, J.H., Oppenheimer,
Society of America, pp. 115–137. D., Reasenberg, P., Jones, L., Raleig, C.B., Wong, I.G., Scotti, O., Wentworth, C., 1987.
Mercier, J.L., Carey-Gailhardis, E., 1989. Regional state of stress and characteristic fault New evidence on the state of stress of the San Andreas Fault System. Science 238
kinematic instabilities shown by aftershock sequences: the aftershock sequences (4830), 1105–1111. doi:10.1126/science.238.4830.1105.

You might also like