See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/282353139 Effects of Physical Processes and Sampling Resolution on Fault Displacement Versus Length Scaling: The Case... Article in Pure and Applied Geophysics · September 2015 DOI: 10.1007/s00024-015-1172-0 CITATIONS 0 READS 72 6 authors, including: Some of the authors of this publication are also working on these related projects: Identification of complex geobodies in seismic images View project Geology and stratigraphy of Cuba. Deposits of the K-Pg boundary. Fossil record of Cuba View project Shunshan Xu Universidad Nacional Autónoma de México 47 PUBLICATIONS 172 CITATIONS SEE PROFILE S.A. Alaniz-Álvarez Universidad Nacional Autónoma de México 61 PUBLICATIONS 780 CITATIONS SEE PROFILE Jose Grajales-Nishimura Universidad Nacional Autónoma de México 20 PUBLICATIONS 667 CITATIONS SEE PROFILE Luis Velasquillo Instituto Mexicano del Petroleo 21 PUBLICATIONS 119 CITATIONS SEE PROFILE All content following this page was uploaded by Shunshan Xu on 27 October 2015. The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document and are linked to publications on ResearchGate, letting you access and read them immediately.
19
Embed
Effects of Physical Processes and Sampling …terra.geociencias.unam.mx/~alaniz/Xu 2015PAgeoph.pdfEffects of Physical Processes and Sampling Resolution on Fault Displacement Versus
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
lation of fault systems over a long period of time (e.g.
GUDMUNDSSON 1987; GUPTA and SCHOLZ 2000). The
segmented structure of faults is a basic characteristic
of most natural arrays showing fault interactions at
different scales (e.g. PEACOCK and SANDERSON 1994).
Two segment linkage models are proposed in the
literature—the isolated and the coherent models (e.g.
CHILDS et al. 1995; WALSH et al. 2003). The isolated
model shows how over time, initially isolated fault
segments grow by tip propagation and will experi-
ence eventual, incidental, lateral overlap and
interaction. In the coherent model, the kinematically
related segments, initially belonging to the same
structure, link into a single array in the final stage
(e.g. CHILDS et al. 1995). Generally, there are two
ways by which fractures link, through curved hook-
shaped fractures (mostly extension fractures), or
through connecting transfer fractures (e.g. FERRILL
et al. 1999; GUDMUNDSSON 2011).
For fault segments in the relay zone, the reorien-
tation, or tilt of bedding, transfers displacement
between the segments. This may produce higher
Dmax/L ratios when one segment enters the relay zone
(e.g. DAWERS and ANDERS 1995; MANIGHETTI et al.
2001; WALSH et al. 2003). However, closer to the
fault tips in the relay zone, the displacement gradient
decreases (WOJTAL 1994; WILLEMSE et al. 1996).
Generally, the Dmax/L ratio increases in the relay
segment for the isolated model (Fig. 1a). This
increase in Dmax/L ratio is due to a smaller segment
length and a greater maximum displacement than in
the case of isolated faults, and is based on the
S. Xu et al. Pure Appl. Geophys.
displacement transfer by a local perturbation of the
stress field, which diminishes the growth of tip zones
near the relay zone (e.g. WILLEMSE et al. 1996;
Fig. 1a). In the displacement–length plot (Fig. 1b),
the position of each segment is located to the left
compared to isolated faults. In this sense, the
exponent deviates from that of an isolated fault with
n\ 1 or n[ 1, depending on the degree of fault
interaction—as relates to spacing and the amount of
fault overlap—for each segment. This case is differ-
ent from fault arrays associated by fault linkage, for
which the position of the total fault array moves to
the right. For isolated soft linkage, if the segments are
taken as isolated faults, the position of each segment
in the Dmax–L plot moves to the left, whereas a
physical linkage results in the position of the fault
array shifting to the right (Fig. 1b, c). For both cases,
the Dmax–L exponent can be less than 1 or greater
than 1.
2.2. Mechanical Layering
GROSS (1993) defines a mechanical layer as ‘‘…a
unit of rock that behaves homogenously in response
to an applied stress and whose boundaries are
located where changes in lithology mark contrasts in
mechanical properties’’. The mechanical layering of
host rocks has considerable effects on the develop-
ment of faults (e.g. BENEDICTO et al. 2003; SOLIVA
and BENEDICTO 2005). Whether a propagating frac-
ture becomes arrested by a layer interface or
penetrates a layer interface is determined by three
related parameters: the induced tensile stress ahead
of the propagating fracture tip; the rotation of the
principal stresses at the interface and the material
toughness or critical energy release rate of the
interface in relation to that of the adjacent rock
layers (GUDMUNDSSON et al. 2010). The mechanism
of a bedding-parallel slip may play an important
role in transferring and accommodating slips within
fault zones that cut across heterogeneous stratigra-
phy (e.g. GROSS et al. 1997; NEMSER and COWAN
2009). In large-scale extensional systems, fault
blocks rotate with progressive extension and bed-
ding rotates to steeper dips. The tendency for
slipping on bedding increases with increasing
extension and block rotation. The actual occurrence
of slips on a bedding surface depends on the
frictional resistance to sliding and cohesion on the
surface. Weak horizons may slip or shear at
relatively low slip tendencies (e.g. FERRILL et al.
1998; ALANIZ-ALVAREZ et al. 1998).
A four-stage conceptual growth model due to the
effect of lithological contacts is illustrated in Fig. 2.
(a)
AB
Dmx
Distance
Dis
plac
emen
tIsolated fault
Fault segment
(b)
Length
Dis
plac
emen
t
n = 1
n 1n 1
(c)
Length
Dis
plac
emen
t
n = 1
n 1
n 1
Figure 1aMaximum displacement increases and fault length decreases for a
fault segment. For isolated faults, linear Dmax–L scaling is assumed.
b Dmax–L data points move towards the left for soft-linked
segments. For isolated faults, linear Dmax–L scaling is assumed.
c Dmax–L data points move towards the right for hard-linked fault
arrays. In both cases, the power-law exponents are less than 1 or
greater than 1, depending on the distribution of the points (hollow
squares or hollow circles). Note that for isolated faults, linear
Dmax–L scaling is assumed
Effect of Physical Processes and Sampling Resolution on Fault Displacement
At stage 1, the fault vertical dimension (H) is less
than the layer thickness, and there is no vertical
restriction. In this case, the fault plane has an ideal
elliptical shape, and the displacement profiles along
both the strike and dip show a triangular shape. This
allows the Dmax–L linear scaling to be ascertained
(e.g. GROSS et al. 1997; SOLIVA and BENEDICTO 2005).
At stage 2, further deformation causes the fault to
approach the interfaces between competent and
incompetent layers, and the fault plane becomes a
quasi-rectangle due to vertical constraints. The fault
growth along the strike is different from that along
the dip. Vertical (dip-parallel) fault growth is char-
acterized by increasing displacement and constant
length, whereas lateral (strike-parallel) fault growth is
characterized by a decreasing Dmax/L ratio. Although
the displacement profiles along the strike and dip are
similar to a mesa (plateau) shape, the mechanism of
displacement accumulation along each is distinct.
Along the fault strike, the displacement accumulation
in the restricted part (central part) is less than that of
areas far from the restricted part (Fig. 2a). Along the
fault dip, the parts near the fault tips are restricted and
displacement accumulation decreases (Fig. 2c). In the
Dmax–L plots, the dip-parallel growth line is vertical
and the strike-parallel growth line is below the linear
line (n\ 1) (Fig. 2b, d; SOLIVA and BENEDICTO, 2005).
At stage 3, a restoring stage occurs when the fault
breaches the mechanical layer. At this stage, fault
growth along the strike follows a constant length
model and the data points in the Dmax–L plot show a
vertical growth path. However, the vertical fault
growth is characterized by a decreasing Dmax/L ratio.
Stage 4 is called the restored stage. In this stage, the
fault plane once again demonstrates an ideal elliptical
shape and linear Dmax–L scaling is expected. The
cycle of this four-stage model may be repeated if the
fault propagates across the mechanical layer bound-
ary and begins to grow within the next larger
mechanical layer.
2.3. Fault Reactivation
Pre-existing faults affect sequent deformation in
two ways. First, the pre-existing faults serve as
nucleation sites for new faults. Second, the pre-
existing faults act as obstacles to the propagation of
the second-phase normal faults (HENZA et al. 2011).
According to the Mohr–Coulomb theory, for a pre-
existing plane, the critical condition for a slip is
(a)
Length
Dis
plac
emen
t
12
3
4
12
3
4
12
23
43
1
23
4(b)
(c)
(d)
Distance
Dis
plac
emen
t
Dis
plac
emen
tD
ispl
acem
ent
Distance
Length
3
4
2
Figure 2a Evolution stage of a fault plane before and after a vertical restriction within a brittle layer. The shaded area indicates the decrease of the
displacement increment at stage 2. b Model of along-strike displacement profiles at different stages. c Along-strike fault growth. The shaded
area indicates the decrease of the displacement increment at stage 2. d Model of vertical displacement profiles at different stages. e Vertical
fault growth
S. Xu et al. Pure Appl. Geophys.
s ¼ C þ lsðrn�PÞ; ð2Þ
where s is the magnitude of shear stress, rn is the
magnitude of normal stress on the pre-existing plane,
C is the shear strength of the pre-existing plane when
rn is zero, ls is the coefficient of friction of the pre-
existing plane and P is fluid pressure (e.g. JAEGER
et al. 2007; GUDMUNDSSON 2011). The value of rn is
dependent on the orientation of the plane relative to
the principal stresses responsible for the reactivation
(e.g. Jaeger and Cook 1969). C and ls depend on
lithology. This means that reactivation processes are
selective and only occur on some portions of faults
(e.g. MORRIS et al. 1996; KELLY et al. 1999; BAUDON
and CARTWRIGHT 2008; LECLERE and FABBRI 2013). If
fault surfaces have a reduced or negligible cohesive
strength, their reactivation is controlled by the coef-
ficient of friction, the state of stress, the fault
orientation and the pore fluid pressure.
Based on the relative slip sense of a reactivation
event on a fault, three types of fault reactivation can
be distinguished, namely, normal reactivation,
reverse reactivation and oblique reactivation
(Fig. 3a). Normal reactivation occurs when the angle
between the new slip sense and previous slip sense
(h) is between 0� and 45�. Reverse reactivation refers
to reactivated faults with an opposite slip
(135�\ h\ 180�) response to changing stress con-
ditions or tectonic settings. When 45�\ h\ 135�,the fault is known as an oblique-reactivated fault
(Fig. 3a). Fault reactivation is an important factor for
modifying fault displacement geometries and for
controlling the pattern of deformation (e.g. CART-
WRIGHT et al. 1995; WALSH et al. 2002; VETEL et al.
2005). A reverse reactivated fault can exhibit a lower
displacement-to-length ratio compared to un-reacti-
vated faults (e.g. PEACOCK 2002; VETEL et al. 2005;
KIM and SANDERSON 2005). A possible result of a
change in the Dmax/L ratio is that the exponent
n becomes less than 1 (Fig. 3b). Normal-reactivated
faults can accumulate more displacement while
maintaining a near constant fault trace length (e.g.
BAUDON and CARTWRIGHT 2008). This disproportion-
ate increase of maximum displacement against length
shifts the growth path in a plot of displacement to
length to a path with a higher Dmax/L ratio (Fig. 3c).
3. Scaling of the Fault Displacement Versus
the Length of the Normal Faults
from the CANTARELL Oilfield in the Southern
Gulf of Mexico
3.1. The Geological Background of the Study Area
The Cantarell oilfield is located in the southern
part of the Gulf of Mexico, 85 km offshore from
Ciudad del Carmen, Yucatan Peninsula (Fig. 4a, b).
The Gulf of Mexico was formed as a result of Middle
Jurassic rifting, which produced passive margins
flanking a small area of oceanic crust in the central
part of the basin (e.g. SAWYER et al. 1991). The
counterclockwise rotation of the Yucatan Peninsula
block away from the North American plate took place
(b)
Length
Dis
plac
emen
t
n = 1
n 1
(c)
Length
Dis
plac
emen
t
n = 1n 1
45°45°
45°45°
Normal
reactivation
Reverse
reactivation
Obliquereactivation
Obliquereactivation
(a)
Figure 3a Classification of fault reactivation. The filled arrow indicates
previous slickenline senses on the fault. The hollow arrows indicate
the reactivated slickenline senses on the fault. b Dmax–L relationship
with a decreasing Dmax–L ratio due to fault reactivation.
c Dmax–L relationship with an increasing Dmax–L ratio due to fault
reactivation
Effect of Physical Processes and Sampling Resolution on Fault Displacement
(a)
(c)
(b)
(d)
(e)
S. Xu et al. Pure Appl. Geophys.
during the formation of the Gulf of Mexico (e.g. Bird
et al. 2005; PINDELL and KENNAN 2009). In Campeche
Bay, three primary superimposed tectonic regimes
are recorded (ANGELES-AQUINO et al. 1994): the
extensional regime initiated in the Middle Jurassic;
the compressional regime during the middle Mio-
cene; and the extensional regime extended
throughout the middle and late Miocene. During the
last two regimes, salt tectonics occurred in all of the
areas, overprinting structures and disturbing the
regional stress field.
Various studies have been published regarding the
structural features of the Cantarell oilfield (e.g.
SANTIAGO and BARO 1992; PEMEX-Exploracion Pro-
duccion 1999; MITRA et al. 2005). The Cantarell Field
has an overthrusted structure and shows an upright
cylindrical fold with gently plunging conical termi-
nations (MANDUJANO and KEPPIE 2006). The western
boundary is a normal fault with a minor strike-slip
component, whereas to the north and east, the field is
limited by reverse faults (Fig. 4d). The oilfield is
composed of a number of sub-fields or fault blocks.
These are the Akal, Chac, Kutz and Nohoch blocks.
The faults in the Akal block are normal faults, but the
observed slickensides in minor faults from the core
samples are generally oblique (XU et al. 2004), which
implies a strike component of displacement on the
faults. Recent interpretation of the geophysical data
suggested that Cantarell is a fold-thrust belt and a
duplex structure related to the Sihil thrust. These
faults in the Cantarell were not formed by simple
shear related to the movement of a larger fault (MITRA
et al. 2005; GARCIA-HERNANDEZ et al. 2005).
The stratigraphic records in this oil field are
shown in Fig. 4e (PEMEX-Exploracion Produccion
1999). The main units include Callovian salt, Oxfor-
dian siliciclastic strata and evaporites, Kimmeridgian
carbonates and terrigenous rocks, Tithonian silty and
bituminous limestone, Cretaceous dolomites, and
dolomitized breccias in the Cretaceous-Tertiary
boundary and Lower Paleocene. The Tertiary system
includes siltstone, sandstone and carbonate rocks.
The producing formation was created when the
Chicxulub meteor impacted the earth (GRAJALES-
NISHIMURA et al. 2000). The upper reservoir is a
brecciated dolomite of the uppermost Cretaceous age.
The breccia is from a shelf failure due to an
underwater landslide when the meteor hits. The
lower producing formation is a Lower Cretaceous
dolomitic limestone.
3.2. Relationship Between Fault Displacement
and Length
For analysis of the relationship between fault
displacement and length, structural contour maps are
used to measure the data of fault displacement and
the fault trace length. Four structural maps of a
1:50,000 scale were selected: the dolomitized brec-
cias located at the top of the Cretaceous/Tertiary
boundary (S1), the top of the Lower Cretaceous (S2),
the top of Tithonian (S3) and the top of Kimmerid-
gian (S4). We measured fault displacements from the
structural contour maps, applying the method pro-
posed by XU et al. (2004). According to this method,
fault vertical displacement (Dv) is related to the
dislocation of contour lines (Dc) across the fault trace
and dip of the corresponding layer (b):
Dv ¼ Dc tan b ð3Þ
To measure the value of Dc, two conditions are
required. First, the contour lines must be approxi-
mately perpendicular to the strike of the fault. If this
is not the case, the value of the bedding dip (b) needsto be corrected (XU et al. 2004). Second, the bedding
dip must not be larger than 35�. The tilts of
stratigraphic units in our study area are consistent
with this condition. For example, the average dip at
the top of S1 is 17.3� (XU et al. 2007).
3.2.1 Effect of Fault Interactions
To study the effect of fault interactions, we analysed
two types of datasets for all of the faults in the studied
area. One type of dataset is from the two-tip faults,
which either do not cut other faults or are not cut by
other faults. Another type of dataset is measured from
the one-tip or no-tip faults, which have branching
Figure 4a Location of the study area. b Sketch map of the Campeche Bay.
c Rose diagram of fault direction in the Campeche Bay. d Structural
contour map of the top of the Cretaceous-Tertiary carbonate
breccias (S1) in the Cantarell oilfield. e Integrated stratigraphic
column in the Cantarell oilfield
b
Effect of Physical Processes and Sampling Resolution on Fault Displacement
faults or are branching faults themselves. The results
(Fig. 5a, b) indicate that the coefficients of determi-
nation (R2) for both linear and power-law
relationships are low, both are less than 0.6. For
two-tip faults, the R2 values for linear regression are
larger than those for power-law analysis (Fig. 5a),
suggesting, even weakly, a linear D–L relationship.
The power-law exponent is approximately equal to
0.5 (n = 0.61), but with a fairly low coefficient
(R2 = 0.42).
However, for one-tip or no-tip faults, the R2 = 0.3
for both linear and power-law relationships is much
lower than the R2 for two-tip faults (R2 & 0.5). The
scaling exponent is 0.49, which is far from the linear
scaling law. Although this low slope is poorly defined