A new method of estimating the ratio between in situ rock stresses and tectonics

based on empirical and probabilistic analyses

L.I. Gonzalez de Vallejo *, T. Hijazo

Department of Geodynamics, Geological Engineering, UniveTSidad Complutense de Madrid, 28040 Madrid, Spain

ABSTRACT

This paper describes a new procedure for assessing the ratio between in situ stresses in rock masses by
means of J( (K-aH Iav, being O"H and av principal stress) and tectonics for purposes of engineering geology
and rock mechanics. The method combines the use of the logic decision tree and the empirical relationship
between the Tectonic Stress Index, ISI, and a series of J( in situ values obtained from an extensive database.
The decision tree considers geological and geophysical factors affecting stress magnitudes both on the
regional and local scale. The IS! index is defined by geological and geomechanical parameters. The method
Keywords:
In situ stress proposed provides an assessment of the magnitude of horizontal stresses of tectonic origin.Results for
Tectonic stress several regions of Europe are presented and the possible applications of the procedure are discussed.
Geomechanical index
Decision tree
Logic tree
Probabilistic methods

1. Introduction measure stress magnitudes such as hydrofracturing, overcoring, or

doorstopper techniques are described by ISRM (2003).
In rock mechanics and engineering geology, certain properties of In this paper, a new procedure is described whereby the value of K
rock masses are generally estimated through empirical relationships. (oH/aY) can be estimated for a given rock mass. The method is based on
However, for in situ stresses, the available empirical relationships do applying the probabilistic decision tree method and the empirical
not allow to estimate stress magnitudes within an acceptable range. In relationship between the TSI (Tectonic Stress Index) and K. The decision
this paper K is defined as the ratio between the major horizontal stress tree considers the geodynamic and geophysical factors that determine
(OH) and the vertical stress (ay) (Good man, 1989), being ay the weight horizontal stress magnitudes on both a regional and local scale and
of overburden. Fig. 1 shows K-depth relationships using the stress results are expressed qualitatively as very high, high, intermediate or
data compiled in this study. Envelope lines obtained from these data low magnitudes along with the possibility of local stress amplification
and those proposed by Hoek and Brown (1980) are also included in effects. The TSI takes into account the geological history of the rod<, its
Fig. 1. Large variation in the value of sigma H at the depths commonly elastic modulus and the maximum lithostatic load. Using a large world
dealt with in engineering is observed. Other methods of estimating database of in situ measurements of K, empirical correlations between K
stresses (Sheorey, 1994) are based on the thermoelastic properties of and TSI have been established. The results of applying these correlations
rocks, but do not consider the main factors affecting the state of stress to a wide range of cases are presented.
of the rock. Approaches such as geological (tectonic structure
analysis), seismic (focal mechanisms) give an estimate of the 2. Factors affecting the state of stress
orientations of stresses but not their magnitude. Indirect estimation
methods include acoustic emission (AE), anelastic strain recovery Any method of evaluating the state of stress for rock mechanics and
(ASR) measurements (ViUaescusa et al., 2002; Un et al., 2003) and engineering geology purposes needs to consider the factors affecting
borehole breakouts and core disking. Available procedures to directly natural stresses, including the origins and the mechanisms that
generate these stresses, as well as their spatial distribution and
magnitude. Table 1 provides a summary of the main models and
hypotheses proposed to explain the origins and formation mechan­
isms of the stresses affecting the Earth's crust or upper elastic
lithosphere.
 K IlcrH+crhI2)lcrv) The distributions and orientations of stresses on continental and
o 2 3 4 5 6 7 8 9 regional scales can be found in the World Stress Map (WSM) (Zoback
et aI., 1989; Reineeker et aI., 2004).
The factors that most affect stress magnitudes are:

Rheological behaviour. On the lithospheric scale, this behaviour

controls the relationship between stress magnitude and depth, and
I'500�----+-0-----4 depends on the heat gradient and on crust composition and
�
15. thickness.
o
� 20CIO����f-----�
Heat flow affects stress magnitudes, such that the greater the heat
• Data
flow, the greater is the amplifying effect of stresses in the most
superficial zone of the lithosphere, in which brittle behaviour
predominates. Conversely, in the lower part in which the
predominating behaviour is ductile, stresses decrease with depth
(Kusznir, 1991). The concept of amplifying effect of stresses due to
Fig. 1. Variations in K with depth based on world data compiled for this study. - Present
build-up of stresses in the upper most elastic part of the
study, - - - Hoek and Brown (1980� For the purpose of comparing the envelopes lines used by
Hoek and Brown and the stress database used in this paper. K values have been plotted as lithosphere. This effect is the consequence of the more ductile
defined by Hoek and Brown: K=(OH+Oh/2)/a... behaviour of the lower lithosphere compared to the upper
lithosphere, which leads to transfer of stresses from the lower to
Tectonic stresses are the main causes of stress in the lithosphere upper area (Kusznir and Bott, 1977). The effect is most intense in
and are generated through two basic mechanisms (Fig. 2): plate margin zones.
Plate boundary forces generated by the movement of tectonic Crust thickness. Since the upper crust layer is mainly comprised of
plates give rise to compressive or extension stresses. These stresses quartz and feldspar and the lower layer of olivine, the uppermost
can reach magnitudes of 50 MPa at collision borders and 20 MPa at portion of the crust is weaker. Thus for high crust thicknesses, the
expansion margins (Park, 1988). proportions of quartz and feldspar will exceed that of olivine and
Forces produced by isostatically isostatically-compensated loads due stress magnitudes will be lower.
to large topographical elevations (mOlllltain ranges) whose weight is Rock composition and its geomechanical behaviour. These two
compensated by zones of less lithospheric density or by an increase or factors are linked: depending on the nature of the rock, its
reduction in crust thickness. This mechanism of isostatic compensa­ geomechanical behaviour will be more brittle or ductile, thus
tion leads to a combined effect of vertical loads and a rising push affecting strength and state of stress.
(buoyancy forces), generating horizontal stresses in adjacent zones. The area affected by plate margin forces depends on the thickness
Their magnitudes can be of the order of 50 MPa (Park, 1988). of the elastic crust enduring these forces. Thus, the greater the
thickness the higher the wavelength or surface affected by the
Both types of stress are permanent and continuous over geological
applied forces.
time and constitute the so-called renewable stresses. Coexisting with
Time of stress. As the time of stress increases, so does the
these stresses are those denoted non-renewable.These are not tectonically
amplification effect along with the thickness of the elastic crust.
significant since they are not long-standing, being gradually released over
time. However, they do give rise to brittle fractures and creep processes. However, due to this increased thickness, stress magnitudes are
The main non-renewable stresses are (Bott and Kusznir, 1984): lower in intraplate and low heat flow regions (cratons), than in
more tectonically active regions with high heat flows and a thinner
Flexural stresses, due to non-isostatically-compensated loads.
crust (Kusznir and Bott, 1977). In regions in which moderate or low
Membrane stresses, due to changes in the Earth's curvature. stresses are exposed to a prolonged period of tectonic forces, the
Thermal stresses, due to differential heat gradients. effect may be comparable to that experienced by regions of high or
Loading stresses, due to sedimentary processes, piling volcanic
rocks, glacial ice deposition or unloading stresses due to erosion and
ice retreat.
Table 1
The regional distribution of stresses depends on the prevailing Stress models
tectonic regime. Two large types of setting can be distinguished:
Bott and Kusznir (1984}
Intraplate regions in which compressive stresses predominate, Renewable stresses (subjected to amplification) Plate margin forces
which are largely uniform both in terms of their orientation and Forces due to isostatically­
compensated loads
geographical extension. In these intraplate zones, the orientation
Non-renewable stresses (not subjected to F1exural stresses
of the compressive stress field depends of on the following factors: amplification effects) Membrane stresses
compressive plate margin forces, ridge push and continental Thermal stresses
collision stresses as well as the geometry of the plate margins on
Zoback et al (19S9}
which they act. Discrepancies both in magnitudes and orientations
First category stresses Plate margin forces
can be attributed to buoyancy forces (Zoback et al., 1989).
Forces generated by geodynamic
Continental regions with large mountain systems. Here, the processes
predominance of extensive stresses affects the different tectonic Thermoelastic forces
Shear forces at the base of the
settings (continental collision, intraplate rift, back-arc regions). In
Iithosphere
some cases, changes in the direction of extension stresses are Forces arising from plate geometry
related to lateral changes in the thickness of the lithosphere and Second category stresses F1exural stresses
Buoyancy forces
heat flow differences.
 OROGEN

RIDGE PASSIVE
TRENCH
�-----r----__�BORDER RIDGE

1
G
Fig. 2. Locations and relative orientations of forces generating stress fields in the lithosphere: FpT (plate traction): FRP (ridge push): hPT (resistant to the plate): Fep (plate contact):
FIT (transformant faults): FBo (basal drag): hSB (extensive stress linked to lateral density variations and buoyancy forces): he (crust flexion): he (lateral strength differences).
modified from Olaiz et al. (2006).

very high stresses sustained over a shorter time period (Zoback expert evaluates the magnitude of major horizontal stress as a
et aI., 1989). function of the different factors considered in the tree that contribute
Residual stress. Can remain in rock masses which have been to the state of stress. Hence, certain criteria need to first be established
(Table 2) to qualitatively assess these magnitudes (as high, inter­
subject in the past to higher stresses than they are subject to today
mediate or low).
to different conditions, so that allow the rock to reach a new
Like a logic tree, a decision tree is comprised of nodes and
internal equilibrium under reduced load or temperature changes.
branches. Each node represents a variable or state of the process to be
These stresses were found to be relatively small compared to the analysed. Each branch indicates a possible option and is assigned a
high horizontal stresses measured at shallow depth (Amadei and probability value or degree of confidence that the node or variable is
Stephansson, 1997). correct. All branches lead to nodes that are true states, and their
In summary, the factors that mostly affect the magnitude of natural probability value depends on the degree of confidence or knowledge
stresses are: crust thickness, heat flow, time over which stresses are of the question considered. Once these probability values have been
sustained, and the composition and heterogeneities of the crust. To a assigned to each branch, all the probabilities of the different branches
similar extent as crust thickness, the predominant factor is heat flow arising from the same node should add up to 1. The final node is
and therefore the value of UH is greater in regions of high heat flow. In assigned a percentage or degree of confidence based on the relative
contrast, like the heat flow, crust thickness controls magnitude such influence of the different contributing factors.
To use this method to estimate horizontal stress magnitudes, the
that lower UH values are observed for great thicknesses, as it has been
main factors affecting the state of stress described in Section 2 were
previously explained.
included. To this end, two decision trees were constructed. The first
Local factors can also substantially modify both the magnitude and
(tree 1, Fig. 3) included regional factors affecting horizontal stresses.
orientation of stresses. These include: rock composition, faults or
Its results indicate whether stresses are low, intermediate or high. For
fractures, sedimentary loads, topographic effects and glacial
unloading. the present purposes, stresses were defined as low if their magnitude
Fig. 1 illustrates how stress magnitudes change with depth, and was less than 10 MPa, intermediate if it was 10 to 25 MPa, high if 25 to
shows the variation in K with depth for the data set recompiled for the
present study. The figure indicates the lower and upper limits of K Table 2
according to depth as reported by Hoek and Brown (1980), which may Factors contributing to the state of stress considered in the decision tree to estimate
horizontal stresses
be compared with the present limits. In both cases, the value of K is
clearly greater than 1 and may even exceed 6 for the first 500 m of Factors Magnitude of predicted horizontal stresses
depth, dropping to 2 beyond depths of 1000 m and tending towards High Intermediate Low
values close to 1 at depths of 2000 to 3000 m, despite few K Geodynamic Intraplate (non Intraplate (cratons)
measurements available for these depths. setting cratonized regions)
Convergent plate Convergent plate margin Divergent plate
3. Use of the decision tree to estimate horizontal margin and transforming fault margin

stress magnitudes
Heat flow High Intermediate Low

Crust thickness Low Intermediate High

The decision tree procedure is based on the logic tree method that
Geomechanical Fragile. high Fragile. moderate Ductile. low strength
is widely used in engineering and well known in seismic hazard behaviour strength and strength. resistant and and low elasticity
analysis (Coppersmith and Young, 1986; EPRI, 1987; Reiter, 1990; US elastic elastic
Army Corps of Engineers Internet Publishing Group, 1999). It is a tool Tectonic Compressive Compressive and strike- Extensive
for helping to choose between several course of action. It provides a structures slip
Topographic Very high relief or High to moderate relief No topographic
highly effective structure within which it can be laid out options and effect abrupt relief effect
investigate the possible outcomes of choosing those options. The Preexisting High erosion Moderate erosion No appreciable effect
decision tree estimates the likelihood that the particular question overloads
addressed will be true. Its results represent a degree of confidence in Tectonized Very tectonized Predominant fractures Absence of or few
zones fractures
the prediction rather than a probability (Bommer et al., 2005). The
 GEODYNAMIC Geomechanical behaviour. This factor was considered at the
regional scale according to the brittle or ductile behaviour of the
materials. If the materials are brittle, the tectonic setting will be
characterized by fractures or normal or inverse faults; in ductile
materials, the tectonic environment will be characterized by folds,
which are common in sedimentary settings.
Distensive For tree 2, the following factors were considered:
eratons.
Low heat
Topographical effects: these occur in regions with large elevations
flow and Thrusts
and deep valleys. In areas of marked relief the orientations and
low/medium
magnitudes of the main stresses are modified.
crust thickness
Cornpressive Preexisting overloads: loads of sediments or of layers of ice of
2
-" glacial origin; erosion or ice retreat increase the relative value of
0-

� No craton
regions.
Distensive
horizontal stresses.
Presence of faults and fractured zones: if the region analysed occurs
Medium heat in an area tectonized by faults and other tectonic structures or is close
Thrusts
flow and
to this region, horizontal stresses may be locally amplified.
medium crust
thickness COl11pressive Table 2 shows the factors included in the trees used to estimate the
Maximum relative magnitudes of the horizontal stresses that might be expected.
horz
i ontal
stress Divergent 4. Estimation of f( from lSl
plate Distensive
border. High The TSI was defined according to the expression (Gonzalez de
heat flow and Vallejo et aI., 1988) L
� low crust Thrusts
v
"E
thickness TSI � log(T/(E,H) ,Ne, se (1 )
0
.L
v Convergent
;; Distensive where
c: plate border.
High/medium T age of the first orogenic cycle or main folding period
Thrusts
heal flow and affecting the rock mass (years)
high/medium E elastic modulus of the rock matrix (Gra)
crust thickness Compressive H maximum lithostatic load supported throughout its geolo­
gical history (metres)
Ne coefficient of seismotectonic activity
Fig. 3. Decision tree 1: estimating the magnitudes of horizontal stresses due to regional se coefficient of topographic influence.
contributing factors.
The age (T) of the main folding period or orogenic cycle is related to
40 MPa and very high if greater than 40 MPa; always for depths of less the time elapsed since the main deformation of the rock was
than 2000 ffi. In the second tree (tree 2, Fig. 4), local factors were produced. In effect, this time constitutes one of the factors contribut­
considered, that is, factors that could amplify horizontal stresses ing to the stress state as mentioned above in Section 2. For the TSI, it is
through local effects. The degree of confidence adopted for this assumed that the main orogenic period corresponds to peak tectonic
analysis was at least 60%. stresses and that subsequent to this deformation, tectonic stresses
The following factors were considered for tree 1: diminish with time. In the TSI, T is referred here to the orogeny of the
Hercynian (250-300 M,a,) and Alpine (10-12 M,a,), although we also
Geodynamic setting: intraplate or plate margin regions. For the
considered the Caledonian orogeny (600 M.a.). In the situation of a
purpose of this study, a plate margin region was defined as one
single rock mass being affected by several orogenies, only the most
whose deformation is directly associated with the actions of the ancient of these was considered.
plate margin. Examples of this type of deformation are the large The variable E is related to the petrophysical properties of the rocks
orogens of the Alps, Andes, Himalayas or areas close to a rift or along with processes of diagenesis, compacting, lithification, defor­
dorsal. mation and recrystallization. High values of E generally indicate
Crust thickness: the following intervals were established: highly resistant rocks able to sustain high stresses before they deform
High crust thickness: 2:30 km (orogen and cratons). or break. Values of E lower than 25 GPa are not valid for applying the
Intermediate crust thickness: 15 to 30 km (intraplate regions). TSI. Notwithstanding, if the same rock mass showed several
lithologies, the highest E value was considered.
Low crust thickness: <15 km (generally the ocean crust).
The variable H indicates the maximum lithostatic load to which
Heat flow: the ranges established were:
the rock has been subjected over its geological history. For
High heat flow: 2:80 mW m-2 (regions close to a plate margin).
sedimentary rocks, H is equivalent to the maximum thickness of
Intermediate heat flow: 45 to 80 mW m-2 (intraplate regions with sedimentary deposits within an error range lower than ±500 m. A
recent tectonics). rough estimate of H can be made from the peak heights of the
Low heat flow: <45 mW m-2 (cratonized regions).
Tectonic structures. The structures considered here were of the
1 The expression defined in reference Gonzalez de Vallejo. 1988 was: log SRF=
compressive type: folds, inverse fractures and overthrusts. Exten­ (T I(E·H))·NC- se (SRF = stress relief factor). In this paper SRF has been substituted by
sional structures were taken to be normal faults. TSI (tectonic stress index) which better describes this parameter.
 4,S
sedimentary setting. Moreover, the tectonic structure in which the
zone examined is emplaced should be taken into account, since in
4,0
•
regions with overthrusts or inverted folds, part of the stratigraphic
sequence may be repeated, increasing the value of H. \
For igneous rocks, H corresponds to the depth at which the rocks
were emplaced and acquired their elastic properties. This depth is
estimated within a margin of error of less than ±1000 m. For extrusive,
or volcanic igneous rocks, H is determined as for sedimentary rocks,
3,5

3,0
\

-\
\

\ 0
\\ •

Z 2,5 -
that is, calculating the current stratigraphic column taking into
account possible eroded levels. For intrusive, or plutonic, rocks, H is
'�
.£ 2,0
'10
0
1
,-

�\.
estimated according to the emplacement depth of the intrusive body. "
\0
This depth may be inferred from several criteria such as texture, grain '\
1,5
size, contacts between the wallrock and intrusive rock; three
\0 • •
emplacement depth levels being differentiated (Buddington, 1959):
1,0 �
\0 \
Epizone: depth less than 8 km. A vast contrast between the
0 • • i'
0,5
temperature of the magma and surrounding wallrock gives rise to 0 •
rapid crystallization and small grain sizes. The wallrock is usually
0,0
affected by contact metamorphism. Contacts between the pluton 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5
and wallrock are angular and discordant, and the high temperature TSI
difference gives rise to porphyric and aplitic textures.
Fig. 5. Relationship between in situ determined K values and the TSI using the T51'88
Mesozone: In the middle crust at depths of 8-12 km. Contacts are database (Gonzalez de Val1ejo et al., 1988). • Hercynian rocks; 0 Alpine rocks; -
angular and discordant as well as gradational and concordant. global relationship for Hercynian rocks; - - global relationship for Alpine rocks.
Foliation in plutons is scarce and often chemically and miner­
ogically zoned. The metamorphic contact zone is extensive and the
at which they formed can be estimated. As a simplification, three
pluton shows flow structures.
zones related to the degree of metamorphism can be distinguished:
Catazone: Emplaced at depths greater than 12 km. The temperature
difference between the magma and wallrod< is small and this gives rise Low-grade metamorphism: temperatures between 200 and 450°C,
to large crystal sizes. The wallrod< rock is of high-grade metamorph­ depths between 5000 and 7000 m. The most common minerals are
ism. Contacts between the pluton and wallrod< are concordant and talc, epidote, chlorite and serpentine. Characteristic rocks: quart­
gradational. Plutons generally show a foliation pattern consistent with zites, slatey granites, milonitized granites, phylites and slates.
that of adjacent metamorphic rocks. Migmatites appear. Intermediate-grade metamorphism: temperatures between 450 and
650°C, depths between 7000 and 12 000 m. The most common min­
For metamorphic rocks, H can be estimated from the minerals erals are biotite, hornblende, staurolite and andalusite. Characteristic
comprising the rock as pressure and temperature indicators (geoba­ rod<s: quartzites, micaceous slates, marbles, amphibolites and eclogites.
rometers and geothermometers), and from these variables the depth High-grade metamorphism: temperatures between 650 and
fusion, depths between 12000 and 20000 m. The most common
INFLUENCE minerals are sillimanite, forsterite, wollastonite, garnierite, olivine
Tectonized Stress
FACTORS Topogra- Preexis- and garnet. Characteristic rocks: orthogneisses, paragneisses,
zones present amplifi -
AT LOCAL phical Icnl loads
granulites, graphites and hornblende gneisses.
SCALE effects In the area cation

For the TSI, Eq. (1) includes the coefficients NC and se. NC is the
Faults or
coefficient of seismotectonic activity and was fixed at 0.25 for zones close
tectonized zones
to seismogenetic faults or to plate boundaries; its value is 1 in the absence
Yes
of these conditions. The Se, or coefficient of topographic influence, was
Non fractures zones
fixed at 0.3 for zones close to steep slopes or deep valleys, and at 1 when
Yes
these conditions were not met. In cases where both NC and SC are
Faults or
applicable only the 0.25 value must be considered. These coefficients
tectonized zones
Horizontal were estimated from theTSI'88 database (Gonzalez de Vallejo et al., 1988).
No
stress
amp lifi c ation Non fractures zones
5. Relationships between f( and TSI

Faults or
The first relationship between K and the TSI was derived from a
Yes tectonized zones
database of 38 data denoted T51'88 (Gonzalez de Vallejo et aI., 1988)
corresponding to 38 different zones in which several in situ stress mea­
Non fractures zones
No surements were taken to give a mean K value (Fig. 5). The instrumental K
values are the mean values on each site. These relationships were:
Faults or
tectonized zones For rocks deformed during the Hercynian orogeny:
No
Non fractures zones K'88her = - 1.81· TS! + 7.96. (2)

For rocks deformed during the Alpine orogeny:

Fig. 4. Decision tree 2: estimating the likelihood of amplification effects on the
magnitudes of horizontal stresses due to local contributing factors. K'88alp = -2.57 . TS! + 7.38. (3)
 4,5 4,5

"
1 · · 1
. 0

"
4, 0 1--\-0 -0
•

\
4,0
\

·
\
3,5
\

\(
3,5
\>\0 \ 0
3,0
0
•
\ 0
\ 0
•

\ 3,0

\\i'""
"
·1..
0

:E 2,5 \ •

. \ 0 08
� \ " 2,5 (' 0 0
�
� 2,0 1� •

..--.
••

.. .
'i{i
,50
\ 6l
"
g
·

\
0
.. . '" 2,0 -0
'\'0

"".�
·

" ,
•

1,5
0
\ •

-
•
·
0
8
\ ., ·
Od'<,,� - -
V. . .
o 0 •
• • 1,5
1,0
o •
�� 0 "
0
le 0 . "
/1,8"\\
0,5 0
\
\
•
\ 1,0
°o't0 " \
0 ·
o
\
0,5 0
0,0 00 "
1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5
TSI 0,0
1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5
Fig. 6. Relationship between in situ-determined K values and the TS! using the TSI'04
database (Hijazo, 2004). • Hercynian rocks; 0 Alpine rocks; - global relationship for TSI
Hercynian rocks; - - global relationship for Alpine rocks.
Fig. 8. Relationship between in situ-determined K values and the TS! using the T51'06
database. • Hercynian rocks; 0 Alpine rocks; - global relationship for Hercynian
rocks; - - global relationship for Alpine rocks.
From a subsequent database, T51'04 (Hijazo, 2004) comprising 74
sites, we obtained the following K-TSI relationships (Fig. 6): (W5M), literature review and underground work projects have been
For rocks deformed during the Hercynian orogeny: analysed (Fig. 7). From these data, 172 were selected on the basis of
their acceptable quality in terms of the techniques used for mag­
K'04her -1.99· TS! + 8.73. nitude measurements and coherency in results, along with geolo­
=
(4)
gical and geomechanical information on the rock mass. Each datum
For rocks deformed during the Alpine orogeny:
corresponds to a different site, the mean value of K being taken as
-2.55 . TS! + 7.23. representative of the given site.
K'04alp = (5)
The K-TSI relationships obtained from the T51'06 database are
By incorporating new stress magnitude data, the T51'06 database provided in Fig. 8. The incorporation ofWSM world data has meant a
was compiled. 13 854 magnitude data from the World Stress Map greater variation of data for the K-TSI relationship, owing both to local

• Data from WSM (Reinecker et al., 2004)

Fig. 7. Locations of stress magnitude measurements in the T51'06 database.

 6,------.---,--�r_, Table 3
Ranges of K and possible tectonic stresses indicated by the TS!

Rocks folded in the Hercynian Rocks folded in the Alpine

5
TSI , Stressb TSI , Stressb
K'0611., K'06alp

>3.80 <1.00 Inw >2.46 <1.00 Low

4+-------�----__67�?�7L----� 3.80-3.55 1.00-1.50 1.00-1.49 Medium 2.46-2.25 1.00-1.49 Medium
3.54-3.30 1.50-2.00 High 2.24-1.99 1.50-2.00 High
<3.30 >2.00 Very high <1.99 >2.00 Very high

• K'06=K estimated from the TS! using the 2006 database.

b Stress indicates tectonic stresses.

2+-----�7'��7L----_+��
_ Deviat���
ion K ! 0 , 5
_ Deviation K!O.7 K). In each case, the complete set of data such that the relationship
1 -- TSI'88 obtained is the global scope has been used.
2 --_. T51'04
According to these results, several intervals of K and associated
3 _.._,,-, T51'06
horizontal tectonic stresses as shown in Table 3 were established. As
most of the database used to correlate K and TSI comes from
O�������
o 2 3 4 5 6 hydro fracturing tests, UH and Uy are considered as principal stresses.
K empirical An estimation of UH can be obtained from K and Uy•

Fig. 9. Linear relationships generated by minimum-square fitting in situ K to empirical 6. Application to several regions of Europe
K using the data from Fig. 8 for Hercynian rocks.
The method described was applied to several regions of Spain,
Germany, France, Italy, Switzerland and Austria. In Spain, 25 zones for
anisotropies and to uncertainties in E and H, whose values were which we have in situ stress magnitude measurements were examined.
indirectly determined. The following K-TSI relationships were Fig. 11 shows the K-TSI relationships obtained, indicating that most K
obtained: values range from 1-2 and that the K-TSI relationship both for
Hercynian and Alpine Spain is similar to the K-TSI relationship obtained
For rod{s deformed during the Hercynian orogeny: for the entire data set (6) and (7). The K-TSI relationships for Spain are:

K'06her = - 1 . 93· TS! + 8.38. (6) For rocks deformed during the Hercynian orogeny:

For rod{s deformed during the Alpine orogeny: KSpainher = - 2.27· TS! + 9.51. (8)

K'06alp = - 2.09· TS! + 6.15. (7) For rocks deformed during the Alpine orogeny:

When the three K-TSl relationships obtained from the three KSpainalp = - 2. 45· TS! + 7.27. (9)
databases were compared, K values derived from the K-TSI'06
If we compare the K values determined in situ (in situ K) with
relationship were found to best fit the K values provided by in situ
those obtained by the TSI (empirical K) using Eqs, (6) and (7), 72% of
determinations, both for the Hercynian and Alpine rocks (Figs. 9
and 10). In other words, the in situ K-empirical K relationship for the
'06 database was closest to the ideal relationship (in situ K empirical ==
4,0

6 3,5 \ A

3 3,0
�
5
"
/ ,
, 6
\
/

I�A\
2 2,5
/
/./ 1 "
6
4 - -

;:/
'i!i
.
.
,
2,0
t:
= /' ,
'iii � '\
3 ". "
S 1,5

;
. 6'
� ,

�
lI"'.
2 " 1,0

,,;j _ Deviation K! 0.5

"

�\
V
_ Deviation K! 0.7
0,5
1 --- T51'88
2 --_. T5t'04
3 -..-..-. T$t'06
0,0
1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5
o
o 2 3 4 5 6 TSI

Kempirical Fig. 11. Regional K-TS! relationships for Spain.... Hercynian rocks (hydrofracturing): •
Hercynian rocks (overcoring): /':" Alpine rocks (hydrofracturing): 0 Alpine rocks
Fig. 10. Linear relationships generated by minimum-square fitting in situ K to empirical (overcoring): - - regional K-TS! relationship for Spanish Hercynian rocks: -
K using the data from Fig. 8 for Alpine rocks. regional K-TS! relationship for Spanish Alpine rocks.
 4,5
Using these data (Vera, 2004; Femandez et al., 1998; Tejero and Ruiz,
0 2002), decision trees 1 and 2 were constructed (see Table 2 and Figs. 3
•
4,0
and 4 for the data used).
\ By comparing the in situ determined horizontal stress magnitudes
3,5 \ with those estimated by the decision trees, out of 11 cases analysed,

'-
for 10 the results of the decision trees were correct and for 9 both the
3,0 decision tree and TSI results were consistent with in situ K

-
\

\\..,,
measurements.
:J

,
'iij
2,5
\
\
0
, For other central and western European areas, stress magnitude
o g 0'
data has been grouped together since these regions show similar
c: \
:.::
stress orientations (G6lke and Coblentz, 1996). These data indicate K

:�
2,0

,
values for Hercynian rocks of 1 to 2 (Fig. 12). The K-TSI relationship
\ 0

,- . ··t
1,5
'" obtained for Hercynian rocks shows a somewhat different relationship
�O .. 0 to that derived from the world data Eq. (6). In contrast, for Alpine rocks
.9,. the relationship obtained was close to that based on world data.
1,0
0'
7. Worked example
0,5 0
0
0
As an example of the use of the decision tree and TSI, a case study
0,0
to determine the stress state of a rock mass for a 4.2 km-road tunnel
0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5
project in NE Spain were was performed (Fig. 13). The materials are
TSI sedimentary rocks (shales, sandstones and conglomerates), and the
maximum overburden of the tunnel is 550 m. The tunnel's geological
Ag. 12. Regional K-TS! relationships for Hercynian and Alpine rocks of western Europe.
... Hercynian rocks (hydrofracturing): • Hercynian rocks (overcoring): • Hercynian location is the southern flank of the great Bellmunt anticline, within
rocks (doorstopper): h. Alpine rocks (hydrofracturing): 0 Alpine rocks (overcoring) and the folded foreland of the Pyrenean Range. Crust thickness in this
o Alpine rocks (doorstopper): - regional K-TSI relationship for western and central region ranges from 30 to 40 km and heat flow from 50 to 70 mW m-2.
European Hercynian rocks; - - regional K-TSI relationship for western and central No great relief differences exist in the area, although some faults that
European Alpine rocks.
could locally modify the state of stress were observed. These data were
used to develop decision trees 1 and 2 shown in Figs. 14 and 15. The
the data show a deviation of K±O.5, while for the regional K-TSI results obtained indicate that foreseeable horizontal stresses are
relationship (8) and (9) this deviation of K±O.5 was shown by 84% of intermediate (10 MPa<uH�25 MPa) at a 63% confidence level. Local
the data. amplification effects on horizontal stresses would be expected at an
The decision tree was used to estimate the magnitude of horizontal 80% level of confidence.
stresses in 11 zones of Spain, for which horizontal stress magnitudes The TSI was calculated using Eq. (1). Twas taken as 12 M.a. given
and the necessary geological and geophysical data were available. these are Tertiary rocks affected by the Alpine orogen. The value of H

Site

SPAIN

o I Km

1: Marls, 2: Sandstones, 3: Sandslones and marls, 4: Shales, sandstones and conglomerates, 5: Sandstones
and marls, 6: Sandstones, 7: Sandstones and conglomerates, 8: Shales. F: Fault

Fig. 13. case study location.

was estimated as 1200 to 1500 m from the stratigraphic columns
INFLUENCE Topogra. Tectonized Stress
inferred from geological maps of the zone and the tunnel's geological Preexis-
FACTORS
cross-section, although we only considered the top limit of this range. phica! zones present amplifi-
AT LOCAL tent loads
The modulus E was obtained by uniaxial compressive strength tests SCALE effects in the area cation
using extensiometric bands at 30 GPa. Incorporating these data in
Eq. (1) gives a TSI =2.42. K was calculated from the K-TSI relationship
for Alpine rocks Eq. (7) rendering K =1.09.
Hydrofracturing tests in the tunnel provided K values in the range Fauh or
lcetonic zones
0.86 to 1.57 for depths of 200 to 440 m, with a representative range of
na
Hori7.0 t l
I
1-1.5. Thus, the K value estimated from the TSI was within the mean stress
range obtained in situ and was also consistent with the decision tree amplification
Non fractures
results. zones
o

8. Conclusions

This paper describes a new method of estimating the magnitude of

horizontal stresses generated by tectonic activity. The method is based
on jointly applying the decision tree probabilistic method and the
Yes=0.8 ·1·1·1 =: 0 .8 =: 80 %
empirical relationship between the TSI index and in situ determined K
No � 0.2 ·1·1·1 �0.2
values comprising an extensive database.
The procedure considers the main factors that affect stresses of Fig. 15. Decision tree 2 for the case study.

tectonic origin. In the decision tree, the factors taken into account are
geodynamic and geophysical factors that affect the state of stress. The
decision tree estimates the magnitudes of horizontal stresses due to
regional stress fields grading them as high, intermediate or low. In
addition, it determines the likelihood of stress amplification due to
local effects. The results are expressed as a probability or degree of
confidence. The TSI index takes into account geological and geome­
chanical factors that may affect stress magnitudes such as the age of
the rock deformation, the elastic modulus of the rock or its maximum
lithostatic load. From the TSI and its relationship with in situ
Cratons. CIOW determined K values, a value of K for the rock mass under study is
Low heat
� M,d
0.7

�I
Brittle obtained.
lo���e���m 0.2 03 13 854 data from the WSM, literature reviews and underground
crust
thickness
Di","'i' HiFh projects have been analysed, and selected those showing stress
0 0.1
�
magnitudes of acceptable quality along with the corresponding
rr=ow
0.8 geological and geomechanical information on the rock mass. This
[Ductile
� Mod analysis rendered 172 data representing a wide geographical
Intraplat 0.8 0.2
I distribution. These data were then used to establish relationships
� between in situ measured K values and values of the TSI index, both
'LoW for data from the same region and for the entire data set.
0.3
� Mod
�I
Brittle Several K-TSI relationships were examined for a regional setting,
No�tons 0.2 0.7 for Europe and also on a world wide basis using the entire data set. For
regions
m;<lium heat
flow and ThM"
0.6
[5E Spain, 84% of the empirical K values estimated from the regional K-TSI
medium crust relationship showed a variation of K±O.5 with respect to K values
thickness 'LoW
0.4
raXimum I measured in situ, while this variation was exhibited by 72% of the K
Ductile �.6d
r horizontal
st ress
[ 0.8 values estimated from the K-TSI'06 global relationship. Similar results
[5E were obtained for the remaining regions examined. Moreover, K-TSI
relationships for the global data set (TS1'06 database) revealed a
'IOW
0.1 deviation lower or equal toK±O.5 for 63% of the data and one of K±0.7
I
Brittle for 74%. The lowest deviations observed corresponded to K-TSI
0.2 �
. 7
Compresslve �
relationships within a single region. This is the consequence of the
03 K greater uniformity of geological conditions and less uncertainties in
'IOW
\ 0.2
Mod
calculating the variables comprising the TSI index. In any case, both
IDuctile
0.8 0.8
regional deviations and, in larger measure, global ones, reflect the
complexity of the factors contributing to the state of stress and the
� hypotheses ofTSI index, along with the uncertainties inherent in the
Plate gordersJ stress measurements themselves.
By applying the decision tree procedure to 11 Spanish zones, the
results for 10 of these zones were consistent with in situ stress
Low magnitude � 1·1 (0.7·0.2·0.1 + 0.8·0.8·0.1+0.3·0.2·0.6+
measurements and for 9 zones, both the decision tree and the K values
0.4-0.8·0.6+ 0.1·0.2·0J+0.2·0.8·0J)� OJ6
determined empirically based on the TSI index were comparable to
Medium magnitude � 1·1 (OJ·0.2·0.1+ 0.2·0.8·0.1+ 0.7·0.2·0.6+
0.6·0.8·0.6+0.7·0.2·0J+0.8·0.8·0.3)� 0.628" 63%
instrumental K measurements.
High magnitude � 1·1·0.2·0.2·0.3 � 0.012 Through the combined use of the decision tree and TSI index it is
possible to estimate the magnitude of major horizontal tectonic
Fig. 14. Decision tree 1 for the case study. stresses in terms of K values. The use of this procedure will also
contribute to interpreting in situ stress measurements and extra­ Gi:ilke. M .. Coblentz. D .. 1996. Origins of the European regional stress field.
Tectonophysics 266. 11-24.
polating them to different zones of the rock mass. Gonzalez deVallejo. LI.. Serrano. A. Capote. R .. De Vicente. G.. 1988. The state of stress in
Spain and its assessment by empirical methods. Procc. Int. ISRM Symposium on
Acknowledgements Rock Mechanics and Power Plants. Balkema. Madrid. pp. 165-172.
Goodman. RE.. 1989. Introduction to Rock Mechanics. 2nd. WHey.
Hijazo. T.. 2004. Empirical methods to estimate natural stress of tectonic origen
The authors are grateful to Prof.John Hudson from Imperial College (in Spanish). MSc. Thesis on Engineering Geology. Universidad Complutense de
of London for his valuable comments and suggestions. Also we thank Madrid.
Hoek. E.. Brown. E.T.. 1980. Underground Excavations in Rock Inst. of Mining and
the suggestions of Prof. C. Li, Trondheim University, Norway; Dr. Y.
Metallurgy. London.
Shang, Engineering Geomechanics Laboratory, Chinese Academy of ISRM. 2003. Special issue of the IjRMMS: rock stress estimation ISRM suggested
Science, Beijing; Dr. Mercedes Ferrer, Geological Survey of Spain IGME; methods and associated supporting papers. Int. J. Rock Mech. Min. Sci. Geomech.
40 (7-8). 955-1276.
Dr. Juan M. Insua, Universidad Complutense de Madrid, Dr. Julian
Kusznir. N.J.. 1991. The distribution of stress with depth in the lithosphere: thermo­
Bommer from Imperial College, London and Dr. JuWin Garda­ rheological and geodynamic constraints. Phil. Trans. R. Soc. Lond. A 337. 95-110.
Mayordomo, IGME, Spain. Special thanks to Sr. Angel Rodrtguez Kusznir. N.J.. Bott. M.H.P.. 1977. Stress concentration in the upper lithosphere caused by
underlying visco-elastic creep. Tectonophysics 43. 247-256.
Soto, In Situ Testing, Madrid who provided hydrofracture test from
Lin. W.. Hirono. T.. Nakamura. T.. Yamamoto. K.. Matsuki. K.. Imamura. T.. Oikawa. y..
Spain. We are very gratefully to the reviewer Prof. Dve Stephansson for Takahashi. M.. Kwasnieswski. M.. 2003. Determination of rock stress by anelastic
many helpful comments on our manuscript. strain recovery measurement of an oriented core in the Nittsu region of Japan.
Proceedings of the 3th International Symposium on Rock Stress RS Kumamoto.
Balkema. Kumamoto Oapan). pp. 225-229.
References Olaiz. A. DeVicente. G.. Munoz-Martin. A. Vegas. R .. 2006. Mapa de esfuerzos de Europa
a partir de mecanismos focales calculados desde el tensor momento sismico.
Geogaceta 40. 55-58.
Amadei. B. and Stephansson. 0.. 1997. Rock stress and its measurement. Ed. Chapman
Park. RG.. 1988. Geological structures and moving plates. Ed. Blackie and Son.
and Hall.
Reinecker.j.. Heidbach. 0.. Tingay. M.. Connolly. P. and Muller. B.. 2004. The 2004 release
Bommer. j.. Scherbaum. E. Bungum. H.. Cotton. E. Sabetta. E. Abrahamson. NA. 2005.
of the World Stress Map (available online at www.world-stress-map.org).
On the use of logic trees for ground-motion prediction equations in seismic-hazard
Reiter. L.. 1990. Earthquake Hazard Analysis. Issues and Insights. Columbia University
analysis. Bulletin of the Seismological Society of America 95 (2). 377-389.
Press. New York
Bott. M.H.P.. Kusznir. N.J.. 1984. The origin of tectonic stress in the lithosphere.
Sheorey. P.R .. 1994. A theory of in-situ stress in isotropic and transversely isotropic rock
Tectonophysics 105. 1-13.
Int. J. Rock Mech. Min. Sci. Geomech. 31 (1). 23-34.
Buddington. AE. 1959. Granite emplacement with special reference to North America.
Tejero. R .. Ruiz. j.. 2002. Thermal and mechanical structure of the central Iberian
Geological Society of America Bulletin 70. 671-747.
Peninsula lithosphere. Tectonophysics 350. 49-62.
Coppersmith. K.J.. Young. R.R .. 1986. Capturing uncertainty in probabilistic seismic
US Army Corps of Engineers Internet Publishing GrouP. 1999. Engineer Manuals
hazard assessments within intraplate environment. Proceedings of the 3rd National
"Response Spectra and Seismic Analysis for Concrete Hydraulic Structures"'. WWW.
Conference on Earthquake Engineering (Charleston) South Carolina. 1. pp. 301-312.
usace.army.milfinetfusace-docsfellg-manual*m.htm.
EPR!. 1987. Seismic hazard methodology for the Central and Eastern United States.
Vera. J.A (Ed.). 2004. Geologia de Espana. SGE-IGME. Madrid. 890 p.
Seismicity OWners Group and Electric Power Research Institute. PI01-38-45-46.
Villaescusa. E.. Seto. M.. Baire. G.. 2002. Stress measurements from oriented core. Int. J.
2256-14. Report NP-472.
Rock Mech. Min. Sci. Geomech. 39 (5). 603-615.
Fernandez. M.. Marzan. I.. Correia. A. Ramalho. E.. 1998. Heat flow. heat production. and
Zoback. M.L.. et al.. 1989. Global patterns of tectonic stress. Nature 341. 291-298.
lithospheric thermal regime in the Iberian Peninsula. Tectonophysics 291. 29-53.