You are on page 1of 10

Te~t~nop~~sic~~ 56 (1979) Tl7-T26 Tl.

7
o Elsevier Scientific Pubfishing Company, Amsterdam - Printed in The Netherlands

Letter Section
~._~_~~_~~~~_-~~__
Determination of the mean principal diiections of stresses for a given fault
population

JACQUES ANGELIER
Laboratoire de Gkodynamique, Mpartement de Ghotectonique, Universit6 de Paris VI,
75230 Paris, CISdex 05 (France]
(Received February 9,1979; revised edition accepted April 19, 1979)

ABSTRACT

Angefier, J., 1979. Determination of the mean principal directions of stresses for a given
fault population. Tectonophysics, 56: T17-T26.

New methods of determination of stress axes, from field measurements of the orientas
tions of faults and slickenslides, are discussed. Results show that the approximation
which assumes that relative motion along any fault is controlled by a single common
mean stress tensor is valid.

INTRODUCTION

Tectonic analysis of fault populations commonly includes an attempt at


determinating the main characteristics of the stress tensor. From a regional
point of view, the most interest~g ch~ac~risti~s are the directions fazi-
muths and inclinations) of the three orthogonal principal stresses, i.e. the
axes of the stress ellipsoid. In the present paper, new methods of determina-
tion of these axes will be described, which were developed while studying
the neotectonic evolution of the Hellenic Arc, where pure faulting prevails
(Angelier, 1978).
Numerous aspects related to faulting will not be discussed, such as the
use of tension gashes, joints, stylolites and Riedel shears (see Badgley,
1965; Goguel, 1965; Ramsay, 1967; Mattauer, 1973; Vialon et al., 1976;
and especially Anderson, 1942 and Price, 1966).
I shall assume that each population of fault measurements corresponds
to a single tectonic event, governed by a single regional stress tensor. As
fault populations generally observed commonly result from several events,
a preliminary distinction between successive events must be done in the
field, on the basis of stratigraphic and structural observations.
T18

PRELIMINARY DATA REDUCTION

Ten to hundred measurements of distinct faults bearing slickenslides are


usually made in a typical site (quarry, trench, cliff). Table I is a simple
example of such field measurements. A preliminary subprogram converts
each original measurement into a set of up to five vectors which corre-
spond to the five axes commonly utilized in focal mechanisms of earth-
quakes (Fig. 1). Other subprograms permit plotting data, as on Fig. 2 (Note
that, in the present paper, all diagrams are Schmidt’s projections of the
lower hemisphere).
These primary diagrams allow a rough determination of the principal
stress axes, based on symmetries within fault populations. This is partic-
ularly useful for conjugate (or quasi-conjugate) fault systems (Anderson,
1942).

TABLE I

Field measurements on faults (Neogene, central Crete, Greece)

1 2 3 45 1 2 3 45

TYMOOI 045 615 80E N VOO3.00 TYM020 049 535 3@+ N


TYMOO2 036 595 80’~’ N TYMO21 It39 47’rl 115 b
~~~003 090 8Oh! 58’J N TYMOZZ 057 45N 116 N
TY~004 052 68N 70W N TYM023 112 74.5 66W N
TYM005 045 63V 7BW N TyMO24 025 42’J 116 N
TYMOO~ 110 8BPI 59’d N TYM025 034 56W 123 N
TY~00-f 074 78N 65W N TYV026 037 77E 50s N
TYMOOR 046 605 BOW N TYMO27 057 615 62W N
TYMO09 077 6IN 86E N TYM028 068 58N 74E N
TYMOlO 067 565 88E N TYM029 061 675 81W N
TYMOll 049 70 5 b9W N TYt.4030 028 58F 50s N
TYtdO31 030 b9E 85N N
T’f~012 036 5@ 132 N
TYM032 042 635 @W N
TYMn13 058 515 79W N
TYMO14 079 625 77W N TYMO?’ 023 b8E E7N N
TYMO15 056 b2N 04W N TY~034 069 48N f35E N
TYMO35 068 69N 88W N
TYr4016 034 61N 74W N
TYM017 034 605 76W N TYMO36 015 68W BON N
TYMOIB 037 635 85W N TYMO37 094 7ON 72W N
TYM019 068 725 63E N TYMO38 087 71N 69E N

Legend of columns: 1 = reference number; 2 = azimuth of fault; 3 = dip of fault; 4 =


pitch (e.g., 80E) or azimuth (e.g., 132) of slickenslides; 5 = sense of relative motion (N
means normal).
T19

Fig. 1. Fault mechanism. Both current seismological and author’s tectonic nomenclatures
are given for each axis or vector. F-fault plane; Af-auxiliary plane; C or n is perpendicular
to F, A or s is slickenslide; B or o is perpendicular to A and C. P (r) and T (n) are “Pres-
sion” and “Tension” axes of focal mechanisms of earthquakes, making 45” angles with
F and Ap Compressional and tensional right dihedra (relatively to focus) in black and
grey, respectively. Here, fault motion is normal-sinistral.

Fig. 2. Plot of measurements of Table I. All fault motions are normal.


T20

PHYSICAL BASIS OF THE METHODS

In conjugate fault systems, the orientations of fault planes as well as


slickenslides are determined by the principal stress axes, because the same
tectonic event induces both formation of faults and movement along them.
However, faulting commonly uses previous discontinuities as older faults,
joints, tension gashes or stratifications. Thus, slickenslides are the only
key to the reconstruction of principal stress axes, within a given framework
of fault planes. In the methods which I shall discuss, I assume that the
distribution of fault planes and the stress are independent.
It is not possible to take into account the modification of the stress in
time and space during a faulting event within a given volume without com-
plex and impractical measurements and computations. I consequently
assume with Carey and Brunier (1974), following Bott (1959), that motion
on each fault is independent and occurs in the shear direction governed by a
single common mean stress deuiator. This hypothesis is obviously an approx-
imation, since stress patterns are complex and motions on faults are not
independent (e.g., Goguel, 1948; Price, 1966); its main justification lies in
the coherent results obtained. As the inspection of results shows, when the
geological setting suggests that the faulting event is unique, the orienta-
tions of the computed shear stress and of the observed slickenslides are close
on each fault. In addition, it is intuitively reasonable to consider that local
stress variations cancel each other so that the computed mean stress deviator
is representative of the stress applied to the studied volume.

PRELIMINARY ANALYSIS

Figure 3 gives the definitions and symbols used in this paper.


2
1‘21

Fig, 4. Angular variation of shear stress with q, when stress axes N, , N,, N, are fixed.
n = perpendicular to fault plane F; T,, = projection of N, on F; 7, = projection of N,
on F. When q goes from 0 to 1, the shear stress goes from T,, to 7, , travelling an angle
6 (arrows), given by coss = tan (Y, * tan (Ye, where 01, and oig are the angles of F with
N, and N3, respectively, Above, left: normal faulting. Above, right: strike-slip faulting
(here dextral). Below: reverse faulting,

Figure 4 illustrates the variations of predicted slickenslides (i.e., shear


stress 7) on a fault plane, for given axes of principal stresses. The orienta-
tion of 3’ is entirely determined knowing the orientation of the principal
stresses and a number cp which expresses a linear relation between them:

Nz -N3
cp = ----.- (asN, >NNz >N3 withN1 #N3,0<q<1)
N, -N3

GRAPHIC METHOD

If all faults move independently within a single mean stress, $ must be


in the common part of all P-dihedra, while ii;f! is in the common part of all

Fig. 3. Geometry of fault plane and stresses. Roman italics indicate the vectors related
to the geometry of the fault (i.e., measured): the unitary normal of plane F is 2, in upper
hemisphere; the slickenslide on plane F is; (unit vector). f is in upper hemisphere for
a normal fault, in lower hemisphere for a reverse fault. Here, by reference to the hori-
zontal line NH of F, the motion is reverse (and left-lateral); 2, on F, is defined by
2 = 2 At (unit vector).
Greek italics indicate the vectors related to hypothetical stress: the stress vector acting
on the plane F is called 2. It is defined by 2 = T * i? (T is the stress tensor); the projec-
tion of 2 on F (shaded) is the tangential or shear stress 7; w’, on F, is defined by $ =
~~~==~ti\t;i;,onF,isthecomponentof;;and;tparalleltoi;(p=~.;:=7;*;‘).
T22

NOTATION I

Summary of symbols used in text


-.-
-+
“k unitary normal of fault plane number k
-+
Sk unitary stria (slickenslide)

& unitary vector perpendicular to ?t and2 (2 = 2 A 2)


+
Ok stress acting on fault plane
+
lk tangential component of stress (shear stress)
-+
component of stress zk along zk
axes of mean principal stresses

T-dihedra. (see Fig. 1). This is the basis of a simple graphic method (Angelier
and Meehler, 1977).
However, the right dihedra method does not provide a unique determina-
tion of the best mean stress axes because of the non-determination of P,
as it implicitely assumes that cp may vary from 0 to 1 for each fault indepen-
dently. Thus, the unicity of tp is not postulated, and the field data are not
fully exploited. The following methods do completely determine the mean
deviatoric stress tensor T in relative values (i.e., up and directions of stress
axes) by mathematical means.

ITERATIVE METHOD

Carey and Brunier (1974) and Carey (1976) made an iterative search of
T for a population of CKJfaults by minimizing the following function (see
Fig. 3):
k=3c k-v

F1 = c cos’ (&,&) and F2 = C cos2 c;f,&)


k=l k=1

Inversions of &, ?k or 8jj do not change F 1 and F, . In other words, these


m~imiz~tions neglect the sense of relative motions, so that one should
check the polarity of each fault after the determination. I have chosen a
function which depends both on the direction and sense of each fault
motion (Angelier, 1977):

k-1 k--l
T23

Another formula leads to faster compu~tion, while giving similar results:


k=% tk = taII&,?k) if (&,;)k) < 45”
F5=c ti

k=l fk = 1 if 45” < (&,&)< 180”


i

DIRECT INVERSION METHOD

All previous functions were sums of fractional expressions. Therefore,


all derivatives were so complex as to make prohibitive a direct inversion of
the data.
Angelier and Goguel(l978) have proposed such a direct inversion method
based on a least squares minimization of the component of tangential stress
perpendicular to the measured slickenslides, i.e. the component of stress
which would induce a deviation of shear with respect to real slickenslides.
This component is called 3 on Fig. 3, Now, the components of the stress
tensor in the reference system are the following:

The function to minimize is:

s=c Pi
k=l

Cancelling the four partial derivatives of S with reference to d, e, fi 9, re-


spectively, one directly obtains a simple system of four equations, which
provides d, e, f, tan 2JI :
Ad + De + Ef + Gcos$ + &sin+ = 0
Dd + Be + Ff + Iicosll, + KsinJt = 0
Ed + Fe + Cf + Icos$ + Lsin$ = 0
L Mcos29 +iNsin2$ f (Jd +Ke +Lf)cos$ -(Gd +He +If)sin$ =0

The fourteen parameters A to N are simple sums of homogeneous polynoms


(degree 4) of the components of & and & in the reference system. As the
solution is not unique (JI ), the minimum of S must be distinguished among
two extrema. Then, knowing nume~c~ly d, e, f and $I, the principal axes
are easily localized. However, the sign of T must be determined for each
fault; it enables one to distinguish 131 and ga. Of course, the agreement of
T24

all signs must be checked, Finally, the principal stress axes are completely
identified and the value of q is computed (when most faults are conjugate,
the computed value of cp is not very significant, whatever the method. On
the other hand, no method can provide satisfying axes when all measured
faults are geometrically similar).
It must be emphasized that, as for previous methods, a final check of the
computed results is necessary. Table II indicates for each fault measurement
the angle $,?) between theoretical shear stress and real slickenslide (from

TABLE II

Final check list referring to Table I

2 3 5
1101001. .66 -‘18. .63 17. r18
1301002. ~68 -12. .66 .03
1101003. .57 -51. .36 23: .16
1101004. .64 -26. .57 11. .lO
1101005. .67 -15. .64 lll
1101006. .5? -72. .16 ::: l05
1101007. -59 -48. ‘39 22. .15
1301008. .67 -16. .64 4. r04
1301009. .71 .70 I. .09
1101010. l70 -;Y: .68 6. ‘07
1301011. ‘60 -40. .46 14. lll
1301012. .76 10. .75 7. .09
1301013. l74 1. .74 6. .08
1301014. .69 -25. .59 12. .12
1101015. .69 -12. .67 .04
1101016. .67 -12. .65 1:: .15
1301017. .67 -14. .65 7. .07
1301018. .64 -21. .60 2. -.03
1101019. .59 -48. .39 30. .20
1101020. .73 -20 .72 7. .09
1301021. .76 14. .74 15. l19
1101022. .82 21. l77 19. .25
1301023. .s4 -55. l31 31. .16
1301024. ~82 24. 075 5. .06
1301025. l71 -1. l71 5. .06
.55 -56. l31 30. .15
:33::::? .66 -20. .62 23. -24
130102a: .73 -3. .73 19. .24
1301029. r62 -35. .51 .04
1301030. .68 -8. 060 2:: .28
1101031. r59 -33. .50 13. .12
1301032. .64 -22. .60 -.Ol
1101033. .60 -29. .52 1:: .lO
1301034. .81 .77 9. .A2
1101035. .64 -::: .57 -.Ol
1301036. .58 -29. .51 1:: .09
1101037. .63 -26. l57 14. .14
1301038. .63 -29. l55 5. .OS

Legend of columns: 1 = index including ref. number; 2 = modulus of ‘;: (no scale); 3 =
angle (2, F& in degrees; 4 = modulus of 2 (no scale); 5 = angle (7, b), in degrees; 6 =
modulus of P (no scale). Explanation of symbols in Fig. 3.
T25

0 to 180”) and the modulus of the component of stress p’ (see Fig. 3).
This detailed check may show up anomalous measurements (high values),
which may indicate a distinct tectonic event. If justifiable, the elimination
of such anomalous data leads to a new improved computation.
This new method is much faster although the results based on actual
measurements are generally similar, as shown by Fig. 5.

Fig. 5. Mean stress axes computed when using minimizations of F, (iterative method)
and S (direct inversion method). Fault population as on Fig. 2 and Tables I and II,
without weightings.

CONCLUSIONS

The methods discussed give a more complete analysis than previous ones
(e.g. Anderson, 1942). Although they still require major simplifications of
the actual dynamic patterns, these methods, which are based on Bott’s
analysis (1959), satisfactorily account for actual distributions of faults and
slickenslides, within the accuracy of measurements. A meticulous qualitative
structural analysis in the field is indispensable.

ACKNOWLEDGEMENTS

The author thanks F. Combes (University of Orl&ns), J. Goguel (B.R.G.M.,


Paris), G. Guitard, X. Le Pichon and P. Mechler (University of Paris VI) for
pertinent advises on various parts of his study. X. Le Pichon critically read
the manuscript. The research was supported by C.N.R.S. and the University
of Paris.

REFERENCES

Anderson, E.M., 1942. The Dynamics of Faulting. Oliver and Boyd, Edinburgh (2nd
ed., 1951, 206 pp.).
Angelier, J., 1977. La reconstitution dynamique et geometrique de la tectonique de
failles ?l partir de mesures locales (plans de faille, stries, sens de jeu, rejets): quelques
pkisions. C.R. Acad. Sci. Paris, SQr. D, 285: 637-640.
T26

Angelier, J., 1978. Tectonic evolution of the Hellenic arc since the late Miocene. Tectono-
physics, 49: 23-36.
Angelier, J. and Goguel, J., 1978. Sur une m&hode simple de dktermination des axes
principaux des contraintes pour une population de failles. C.R. Acad. Sci., Paris,
SC. D, 288: 307-310.
Angelier, J. and Mechler, P., 1977. Sur une m&hode graphique de recherche des con-
traintes principales (Sgalement utilisable en tectonique et en dismologie: la methode
des dikdres droits. Bull. Sot. Gdol. France, (7), XIX, 6: 1309-1318.
Badgley, P.C., 1965. Structural and Tectonic Principles. Harper, New York, N.Y., 521 pp.
Bott, M.H.P., 1959. The mechanismsof oblique slip faulting. Geol. Mag., 96(2): 109-117.
Carey, E., 1976. Analyse numerique d’un Mod%e mecanique &%mentaire appliqu&
a 1’Etude d’une Population de Failles: Calcul d’un Tenseur moyen des contraintes
?I partir des Stries de Glissement. Thi?se 38 cycle, Tectonique g&r&ale, Paris-&id,
138 pp.
Carey, E. and Brunier, B., 1974. Analyse thdorique et numerique d’un Modble mkca-
nique Umentaire applique 3 1’Btude d’une population de failles. CR. Acad.. Sci.,
Paris, Ser. D, 279: 891-894.
Goguel, J., 1948. Introduction Zrl’btude m&anique des d&formations de l’dcorce ter-
restre. Mhm. Carte ggol., 2nd ed., 530 pp.
Goguel, J., 1965. Trait6 de Tectonique. Masson, Paris, 2nd ed., 457 pp.
Mattauer, M., 1973. Les Ddformations des Mat&iaux de 1’Ecorce terrestre. Hermann,
Paris, 493 pp.
Price, N.J., 1966. Fault and Joint Development in Brittle and Semi-Brittle Rocks. Per-
gamon, London, 176 pp.
Ramsay, J., 1967. Folding and Fracturing of Rocks. McGraw-Hill, New York, N.Y.,
568 pp.
Vialon, P., Ruhland, M. and Grolier, J., 1976. Elements de Tectonique andytique.
Masson, Paris, 118 pp.

You might also like