Downloaded from http://sp.lyellcollection.

org/ at University College Dublin on March 7, 2016

The geometry of branch lines

GRAHAM YIELDING
Badley Geoscience Ltd, North Beck House, North Beck Lane, Hundleby,
Spilsby, Lincolnshire PE23 5NB, UK
(e-mail: [email protected])

Abstract: A branch line is the line of intersection between two hard-linked fault planes, or
between two parts of a single fault plane of more complex geometry. Of interest is whether they
provide any information about the kinematic development of the fault system to which they belong.
Analysis of branch lines from a variety of normal fault networks, interpreted on seismic reflection
datasets, shows that the branch lines are generally aligned parallel to the extension direction.
This relationship is shown to be a feature of polymodal (orthorhombic) fault systems produced
by three-dimensional strain. Branch lines between bimodal faults (conjugate, with opposing dip)
tend to be perpendicular to the slip direction.

The motivation for this short contribution comes than a 3D one (Krantz et al. 2014). The verb ‘to
from a conversation about fault populations between map’ describes the process of recording the geome-
the author and Juan Watterson, in which it was try of a structure, by producing a result (a map) that
asserted that ‘the orientation of branch lines must is 2D and horizontal. There is no 3D equivalent,
tell us something about the direction of slip on the except, perhaps, the cumbersome ‘map out in 3D’
faults’ (Watterson pers. comm. 1989). However, or ‘build a 3D model’. Determining the geometry of
after some 10 years of further research into fault branch lines requires 3D information. This might
geometries and displacements with the Fault Analy- either be assembled from (i) a succession of 2D
sis Group at Liverpool University, the opposite slices or (ii) from multiple intersections with a
assertion was claimed: ‘Branch-line orientations high-relief topography, or, preferably, by (iii) true
are . . . not simply related to the fault slip direction’ 3D imaging.
(Walsh et al. 1999). In this contribution, it is shown An example of 2D slicing to determine 3D fault-
that both of these assertions can in fact be true, network geometry is given by Kristensen et al.
depending on the nature of the three-dimensional (2008), who excavated serial cross-sections across
(3D) strain and the symmetry of the fault system. minor faults within soft sediments. Fault displace-
A branch line is the line of join between two ments were 10 cm or less. To capture details such
faults (e.g. Boyer & Elliott 1982) and, hence, the as relay zones and fault-bound lenses, sections were
branch lines are a fundamental part of the overall spaced at 0.5–2.5 cm intervals. This allowed branch
fault network. However, very little has been docu- points to be correlated from one section to the next,
mented in terms of the observed geometry of branch to build up the geometry of individual branch lines.
lines. Branch lines are not explicit in conventional On a larger scale, 2D slices may be assembled in
cross-section or map representations of structure. the subhorizontal direction – for example, a succes-
Branch points are seen. It is, therefore, appropriate sion of structure maps of different stratigraphic
to first consider how we can determine the orien- boundaries – within the same area. An example of
tation of branch lines in different kinds of datasets. this style of data is given later, from the Gullfaks
A number of different examples of branch lines oil field (Fossen & Rørnes 1996; Fossen & Hest-
are then presented. Finally, the relationship of hammer 1998).
branch-line orientations to the extension direction In a 3D seismic reflection dataset, a more con-
is summarized in the context of the geometrical tinuous data volume is available. However, the
properties of the fault network. conventional approach to interpreting subsurface
geometry in such a dataset is still basically 2D. To
build a 3D model of the fault network, the following
How can we ‘map’ branch lines? workflow is common:
The inverted commas in this section heading are (1) Interpret fault traces (segments or ‘sticks’) on
used deliberately to point out that, even in geology, seismic sections, preferably in two or more
much of our language is based on a 2D view rather directions.

From: Childs, C., Holdsworth, R. E., Jackson, C. A.-L., Manzocchi, T., Walsh, J. J. & Yielding, G. (eds)
The Geometry and Growth of Normal Faults. Geological Society, London, Special Publications, 439,
http://doi.org/10.1144/SP439.1
# 2016 The Author(s). Published by The Geological Society of London. All rights reserved.
For permissions: http://www.geolsoc.org.uk/permissions. Publishing disclaimer: www.geolsoc.org.uk/pub_ethics
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

G. YIELDING

Fig. 1. Illustration of the technique of seismic fault slicing in a 3D seismic reflection dataset (West Cameron). On
the left, splay and master fault planes are shown in blue and pink, respectively. Slice planes are shown on each side
of the master fault, shifted 25 m into the footwall and hanging wall. On the right, the master fault is shown rendered
with the seismic data from the footwall slice. The discontinuity visible in the seismic reflections (arrowed)
corresponds to the branch line where the splay joins the master fault.

Fig. 2. (a) Google Earth view showing location of the West Cameron seismic survey in the northern Gulf of
Mexico. (b) TWT map of the Top Miocene surface in the West Cameron dataset (red denotes shallow; blue denotes
deep). (c) Dip separations (red ¼ 100 m, purple ¼ 0 m) on fault surfaces near the Top Miocene horizon.
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

GEOMETRY OF BRANCH LINES

surface is located (Fig. 1). In the figured example,

the seismic slice shows the data in the footwall
of the master fault, and in doing so provides a
view of the splay fault as it approaches the master;
the splay is visible as an aligned discontinuity in
the reflections in the slice. In the limit of the fault
slice being very close to the master, this discontinu-
ity is equivalent to a view of the branch line. In most
3D seismic datasets, even of mediocre quality, this
technique provides a much more accurate definition
of the branch-line location and geometry.

Example 1: Gulf of Mexico

The Gulf of Mexico continental shelf is character-
ized by listric normal faults driven by the gravi-
tational collapse of the margin (e.g. Diegel et al.
1995). The West Cameron sector is located off-
shore Louisiana, near the Texas border (Fig. 2).
A 3D seismic reflection dataset was acquired by
GECO in 1987 about 40 km offshore, and is widely
used for research and teaching (e.g. Reymond
1994). Miocene –Pleistocene shelf sequences are
cut by an array of mainly SSE-dipping listric normal
faults. Observations of channel displacements con-
firm that the faults have normal-sense offset with
no strike-slip components (Reymond 1994). The
down-to-south sense of the faulting is indicated
Fig. 3. (a) Interpreted fault network in the West by the vertical offsets in an example stratigraphic
Cameron 3D seismic dataset, viewed from the SW.
One seismic row is shown at the eastern edge of the
surface (Fig. 2b), and the dip separations at a compa-
survey. The red arrow indicates the approximate rable level in the fault network are shown in the map
position of the Top Miocene reflection (shown in of Figure 2c.
Fig. 2b). (b) Same view as (a) but with fault surfaces A 3D view of the fault network is shown
removed to show the branch lines. v.e., in Figure 3a. The seismic interpretation of the
vertical exaggeration. faults was mainly performed on a subset of the sur-
vey rows and columns, at a line spacing of 125 m
throughout the survey volume (Kneen 2007).
(2) Model the fault raw data (segments) into fault
These interpreted fault segments were correlated
planes (triangulated or gridded surfaces).
into surfaces by triangulation and gridding. Links
(3) Establish linkages (branch lines) between
between fault surfaces (i.e. the branch lines) were
individual fault surfaces.
constructed using seismic fault slicing – in fact,
The last step, closing the gap between one inter- the example shown in Figure 1 is from the West
preted fault surface and another, is typically handled Cameron dataset. Figure 3b shows the branch lines
in 3D-modelling software in a purely geometri- with the fault surfaces stripped away. In concert
cal way. The ‘splay’ fault surface is extrapolated with the dominant southerly dip of the fault sur-
onto the ‘master’ fault, based on its 3D trend as faces, the branch lines also mainly dip south.
it approaches the master fault. Unfortunately, faults In Figure 4a, the branch lines are displayed in
often curve as they approach and interact (e.g. Segall map view, and show a strong north–south align-
& Pollard 1980), and so, potentially, the resulting ment, with a mean trend of 1768 and mean angular
intersection may be poorly located by using planar deviation of 32.98 (Fig. 4b). In qualitative terms,
extrapolation. this is perpendicular to the approximately east –
A more data-driven way to obtain the branch west trend of the margin (Fig. 2), and is therefore
line in 3D seismic reflection datasets is a technique parallel to the expected transport direction, driven
termed ‘seismic fault slicing’ (Brown et al. 1987). by gravitational collapse into the deep-water Gulf
The 3D geometry of the master fault is used to sam- of Mexico.
ple the seismic data immediately adjacent to the Although the original seismic volume is in the
fault plane, on the side of the fault where the splay two-way-time (TWT) domain, interpreted structures
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

G. YIELDING

Fig. 4. (a) Map view of West Cameron branch lines of Figure 3b. (b) Rose diagram of the branch-line strikes. The
mean strike is 1768 (mean angular deviation is 32.98). (c) Lower-hemisphere equal-area projection of branch-line
trends (average trend per branch line). The mean orientation is a plunge of 47.48 towards 175.68. (d) Principal strain
axes derived from a Kostrov moment summation of the fault network. ]hmax plunges 38 towards 3368 and ]hmin
plunges 0.18 towards 2468.

can be converted into the depth domain using the The eigenvectors of ]ij are the principal axes of
appropriate interval velocity, which is 3 km s21 the strain ellipsoid, and the eigenvalues are the
for the interpretation shown in Figure 3. The true principal extensions. Applying this method at the
dip and plunge of the branch lines can then be com- level of the Top Miocene (Fig. 2) gives the strain
puted, and they are shown in a lower-hemisphere ellipsoid shown in Figure 4d. The maximum exten-
equal-area plot in Figure 4c. Mean plunge is 47.48 sion direction ]hmax is somewhat east of south, with
towards 175.68. an azimuth of 1568/3368. This is 208 away from
An independent estimate of the extension direc- the mean of the branch-line azimuths (Fig. 4b),
tion can also be obtained directly from the fault although it lies within the mean angular deviation
geometry and displacements by Kostrov’s moment (338) from the mean.
summation (Kostrov 1974; England & Molnar
1997). For a sample area A, and taking fault dip
and displacement measurements at an along-strike Further examples: northern North Sea
sampling interval of s, then:
 s  The Viking Graben in the northern North Sea is a
]ij = [(u/(sinu)) (ui nj + ni uj )] north– south-trending rift system that was active
2A in the Permo-Trias and Late Jurassic (e.g. Badley
where u is the fault dip, u is the fault displace- et al. 1988; Yielding et al. 1992; Fraser et al.
ment vector and n is the fault normal vector. 2002). Extension is generally considered to be
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

GEOMETRY OF BRANCH LINES

Fig. 5. (a) Lower-hemisphere equal-area projection of poles to five main faults in the Gullfaks field, from Fossen &
Hesthammer (1998). For each fault, the best-fit circle and its pole are shown; the mean of the five poles plunges 288
towards 0908. Note the cylindrical nature of the fault planes. (b) Depth map of the Statfjord Formation, Gullfaks
field, from Fossen & Hesthammer (1998). The purple lines are branch lines that have been constructed between the
Top Rannoch Formation and the Top Statfjord Formation. Inset shows the location in the North Sea. (c) Rose
diagram of the branch-line strikes in (b).

broadly east– west (e.g. Roberts et al. 1990; Erratt implying that the faults are strongly dip-slip without
et al. 1999), although temporal variations on this an oblique component.
average have also been suggested (Færseth et al. The fault patterns at different horizons published
1997; Davies et al. 2001). Here, we look at three by Fossen & Rørnes (1996) and Fossen & Hestham-
example datasets across the basin, on a transect mer (1998) allow the construction of branch-line
from west to east. All data are taken from the orientations by identifying the same fault intersec-
Lower– Middle Jurassic sedimentary section and, tion on different maps. In Figure 5b, branch lines
therefore, record the Late Jurassic fault movements. determined by this technique are shown superim-
Fossen & Hesthammer (1998) produced a posed on the structure contour map of the Statfjord
detailed structural study of the Gullfaks field on Formation. Depth values at the branch lines are
the western margin of the Viking Graben, including poorly constrained from the contour maps, so only
structure contour maps, stereoplots and graphs. One azimuths are considered rather than dips, and these
of their key observations was that the main faults are plotted in Figure 5c. The mean orientation is
have a cylindical (corrugated) geometry, defined 0938, which is very similar to the slip direction
by local poles to the fault planes plotting along inferred from fault-plane curvature (0908). Hence,
great circles (Fig. 5a). The pole of the great circle the branch lines point down the slip direction.
for each fault defines the fault’s axis of curvature. A more complicated dataset is shown in
Following similar observations by Koestler et al. Figure 6, from the central part of the Viking Graben.
(1992), Fossen & Hesthammer (1998) argued that Interpretation and depth conversion of a 3D seis-
these poles must be close to the mean slip direction mic dataset was used to produce a fault framework.
on these faults. The poles plunge gently eastwards, In Figure 6a, the fault pattern at a pre-rift horizon
with a mean of 288 dip towards 0908 (Fig. 5a). is shown with thin black lines, with coloured
This direction is also down the dip of the faults, lines showing the branch lines extracted from the
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

G. YIELDING

Fig. 6. (a) Fault pattern and extracted branch lines from a 3D seismic survey in the Viking Graben, data courtesy
of StatoilHydro. (b) Lower-hemisphere equal-area projection of branch-line trends (average trend per branch line).
(c) Rose diagram of the branch-line strikes. The mean strike is 0878 (mean angular deviation is 72.28). (d) Principal
strain axes derived from a Kostrov moment summation of the fault network. ]hmax plunges 118 towards 2818 and
]hmin plunges 48 towards 1908.

framework. Although the dominant fault orientation case is computed to be slightly more than 10%
is broadly north –south, there are many subsidiary extension along a trend of 101–2818. Despite there
orientations, leading to the occurrence of fault- being a much greater spread in branch-line ori-
bound compartments. The branch lines show a entations than in the West Cameron dataset, the
similar complexity: their orientations are plotted mean trend of the branch lines, again, has a similar
on a lower-hemisphere projection in Figure 6b, alignment to ]hmax , in this case approximately east –
and show plunges to the east and west as well as west. There is a 148 difference between the branch-
to the north, precluding the calculation of a simple line mean and the maximum horizontal strain
mean. Plotting only the strikes of the branch lines direction.
(Fig. 6c) allows the west- and east-dipping lines to The third North Sea example comes from the
be classed together, and provides a mean strike Troll Field on the eastern margin of the Viking Gra-
of 0878. ben. Fault interpretation by Bretan et al. (2011)
Using the fault framework of the dataset in Fig- was performed on a 3D seismic reflection dataset
ure 6, it is also possible to extract a strain tensor and depth-converted using well interval velocities.
from a Kostrov moment summation, as described Branch-line orientations from the fault framework
earlier, for the West Cameron dataset. This process are shown in map view in Figure 7a, in a lower-
assumes dip-slip movement, as was demonstrated hemisphere projection in Figure 7b and in a rose
above for the Gullfaks dataset. The result is shown diagram in Figure 7c. The fault network is interme-
in Figure 6d, extracted at a mid-Jurassic pre-rift diate in complexity between the previous two exam-
horizon. The maximum horizontal strain in this ples. The mean strike of the branch lines is 0818.
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

GEOMETRY OF BRANCH LINES

Fig. 7. (a) Fault pattern and extracted branch lines from a 3D seismic survey of the Troll field (Bretan et al. 2011).
(b) Lower-hemisphere equal-area projection of branch-line trends (average trend per branch line). (c) Rose diagram
of the branch-line strikes. The mean strike is 0818 (mean angular deviation is 68.38). (d) Principal strain axes
derived from a Kostrov moment summation of the fault network. ]hmax plunges 68 towards 2638 and ]hmin plunges
18 towards 3538.

Calculation of regional strain from a Kostrov Branch-line orientation and extension

moment summation, at a Jurassic pre-rift hori- direction
zon, indicates maximum horizontal extension at
an azimuth of 083– 2638 (Fig. 7d). Hence, branch Although they can provide an excellent picture of
lines and ]hmax are again both aligned approxi- subsurface 3D geometry, seismic reflection data
mately east –west (28 difference), perpendicular do not, generally, give precise kinematic infor-
to the graben axis. mation, in terms of slip vectors. Channel offsets
Branch lines from the three North Sea datasets in the West Cameron dataset, described above,
are shown together in Figure 8, along with the imply approximate dip-slip movement. In the Gull-
corresponding slip direction or strain direction faks dataset, slip is inferred to be down the fault-
indicators. Both the branch-line observations and surface corrugations.
the slip/strain indicators suggest approximately The existence of branch lines requires faults
east –west extension across the basin, with a spread of different orientations (i.e. non-parallel linked
of 128 for the branch lines and 188 for the slip/ faults). Such fault networks are variously refer-
strain indicators. red to as orthorhombic, polymodal or multimodal
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

G. YIELDING

Fig. 8. Synthesis of the three North Sea datasets shown in Figures 5, 6 & 7 (a, b and c, respectively). The upper
row shows rose diagrams of the branch-line strikes. The lower row shows the sense of slip (a) or extension direction
(b & c), expressed in the range 08 –1808.

(Oertel 1965; Reches 1978; Aydin & Reches 1982; principal strains (Krantz 1988). For orthorhombic
Groshong 1988; Krantz 1988; Healy et al. 2006) normal faults, displacement on each set is approxi-
(see Fig. 9a). They result from 3D (poly-axial) mately dip-slip (Oertel 1965). The faults are
strain. Truly orthorhombic systems have four orien- arranged symmetrically about the strain axes and,
tations of coeval faults, arranged as two sets, each of therefore, there are four possible fault-intersection
which includes opposing dips. The angle between directions (Fig. 9a). Depending on the lateral extent
the two sets is dependent upon the ratio of the of the fault surfaces, the intersection lines can be

Fig. 9. (a) Block diagram and stereonet of orthorhombic normal faults, from Krantz (1988). On the stereonet, the
red symbols mark intersections between faults dipping in the same direction; the grey symbols mark intersections
between faults dipping in opposite directions. (b) Perspective view of two faults (same dip direction) intersecting at
a branch line, each with their own slip vector (black arrow). The red arrow represents the local realignment of slip
parallel to the branch line.
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

GEOMETRY OF BRANCH LINES

Fig. 10. Cross-section (left) and strike projection view (right) of an antithetic fault terminating downwards at
another (conjugate) fault, taken from Nicol et al. (1996).

either branch lines (one fault terminates at the other) Intersections between faults of different dip
or cross-cutting locations (both fault surfaces exist direction are aligned in the direction of minimum
on both sides of the line). horizontal strain ]y (grey symbols on the stereonet
Intersections between faults of similar dip direc- in Fig. 9a). Since the faults are dipping in opposite
tion are aligned in the direction of maximum hori- directions, there must be significant differences
zontal strain ]x (red symbols in Fig. 9a). These in their slip vectors and these intersections are not
intersection lines are close to the slip vector on kinematically compatible.
both fault surfaces and, therefore, this geometry The Andersonian model of conjugate (bimodal)
can accommodate simultaneous slip on both sur- faults can be regarded as a special case of the ortho-
faces with relatively minimal strain in the rock rhombic (polymodal) symmetry, in which inter-
volume around the branch line. Geomechanical mediate strain is 0 (i.e. plane strain). The ]y fault
modelling (e.g. Maerten 2000) would suggest that, intersections are then horizontal. A common obser-
locally, the slip vectors rotate to be parallel to vation in conjugate fault systems is that one of the
the branch line (Fig. 9b). This kind of branch line two faults terminates downwards at the other, to
was termed a conservative fault junction by King give a synthetic –antithetic pair where the anti-
& Yielding (1984). thetic fault dies out downwards at the larger fault.
Figure 10 shows a classic example of this from
Nicol et al. (1996), from a seismic reflection dataset
in the Timor Sea. The branch point in cross-section
belongs to the subhorizontal branch line in three
dimensions. Throw on the antithetic fault decreases
downwards to the branch line, which in this case is
effectively a tip-line. The throw on the antithetic
fault does not add downwards to the throw on the
main fault because the angle between the vectors
is too large. In contrast to the orthorhombic (poly-
modal) examples discussed above, the branch line
is perpendicular to the down-dip slip vector.
Another special case that can be recognized
is that of internal segmentation within a single
fault surface. This description is scale-dependent,
in that multiple fault surfaces may exist in detail
but, at the broad scale, the fault zone has a common
Fig. 11. Stages of development of fault-bounded lens orientation. The geometry of branch lines between
in a fault zone, taken from Walsh et al. (1999). P is an normal faults and their sub-parallel splays has
initial branch point; branch lines are shown bold. been summarized by Walsh et al. (1999) using 3D
Arrows indicate the direction of fault slip. seismic reflection data, and by Kristensen et al.
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

G. YIELDING

Fig. 12. Summary of the three generalized relationships between the branch-line orientation and slip/extension
direction. Left to right: single fault zone (unimodal), conjugate (bimodal) faults, orthorhombic/polymodal fault
network. The upper row shows the perspective views of principal (blue) and subsidiary (green) faults; arrows
indicate slip vectors (red vector applies at branch line). The lower row shows corresponding stereonets, with red
symbols indicating kinematically compatible branch-line orientations and grey symbols indicating kinematically
incompatible orientations.

(2008) using serial sections in soft sediment. Walsh a conjugate (bimodal) fault pair or an orthor-
et al. (1999) argued that these structures grow by hombic/polymodal fault network. The branch-line
failure of relay zones in a propagating multi-strand loops (single fault zone) can also occur within the
fault surface (Fig. 11). As seen at Figure 11e, the bimodal and polymodal networks.
ultimate branch-line geometry may be a closed Thus, it is possible for a branch line to have any
loop, with parts of the line both parallel and perpen- angular relationship to the slip direction, from paral-
dicular to the down-dip slip vector – the branch-line lel to perpendicular or anywhere in-between. On a
orientations are not simply related to the slip direc- single propagating fault surface, local irregulari-
tion. If angular differences between the different ties leading to tip bifurcation can ultimately gener-
fault strands are small, slip can occur relatively eas- ate closed-loop branch lines, which can remain
ily regardless of the orientation of the branch line. stable if the enclosed lens has acute-angled edges.
However, as slip incompatibility increases, there is The parts of the branch line that are perpendicular
a greater chance that removal of asperities will act and parallel to the slip direction are edge and
to smooth out the geometry and return the system screw dislocations, respectively.
to compatibility (Childs et al. 2009). Where the normal faults have opposed dip
(e.g. in conjugate pairs), the subhorizontal branch
line between them is perpendicular to the fault
Conclusions slip direction, and acts to terminate the slip field
on one of the faults. By contrast, in a polymodal
The different relationships between branch-line ori- fault pattern produced by 3D strain, branch lines
entation and slip direction (or extension direction) between different fault sets of similar dip are
are summarized in Figure 12. They are controlled compatible with the slip vector. Analysis of branch
by whether the normal fault or fault network has lines in a polymodal fault network therefore
one, two or four fault-plane orientations, corre- provides a simple way of estimating the extension
sponding to a single propagating fault (unimodal), direction.
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

GEOMETRY OF BRANCH LINES

The examples and analysis presented here have Cardozo, N. & Allmendinger, R.W. 2013. Spherical
deliberately been chosen from relatively simple projections with OSXStereonet. Computers & Geosci-
settings where the observed offsets on the faults ences, 51, 193– 205, http://doi.org/10.1016/j.cageo.
have been generated in one phase of deforma- 2012.07.021
Childs, C., Manzocchi, T., Walsh, J.J., Bonson, C.G.,
tion involving predominantly dip-slip movement. Nicol, A. & Schöpfer, M.P.J. 2009. A geometric
In more complicated tectonic settings, it should model of fault zone and fault rock thickness variations.
be expected that additional factors could affect the Journal of Structural Geology, 31, 117–127.
relationship between bulk extension and branch-line Davies, R.J., Turner, J.D. & Underhill, J.R. 2001.
geometry. For example, if two separate phases of Sequential dip-slip fault movement during rifting:
deformation occurred with two different exten- a new model for the evolution of the Jurassic trilete
sion directions, then the early fault surfaces are later North Sea rift system. Petroleum Geoscience, 7,
reactivated with oblique movement (e.g. Henza 371– 388, http://doi.org/10.1144/petgeo.7.4.371
et al. 2011). In this case, a moment summation Diegel, F.A., Karlo, J.F., Schuster, D.C., Shoup, R.C.
& Tauvers, P.R. 1995. Cenozoic structural evolution
would need to be performed carefully using the and tectono-stratigraphic framework of the northern
true slip vectors for each phase, which is unlikely Gulf coast continental margin. In: Jackson, M.P.A.,
to be possible in a seismic reflection dataset. Analy- Roberts, D.G. & Snelson, S. (eds) Salt Tectonics:
sis of branch-line geometries in an analogue exper- A Global Perspective. American Association of Petro-
iment with two known extension directions (like leum Geologists, Memoirs, 65, 109–151.
that of Henza et al. 2011) might provide an insight England, P. & Molnar, P. 1997. The field of crustal
into whether the branch line orientations have any velocity in Asia calculated from Quaternary rates of
relationship to those two directions, and whether slip on faults. Geophysics Journal International, 130,
they evolve during deformation. A more subtle 551– 582.
Erratt, D., Thomas, G.M. & Wall, G.R.T. 1999. The
effect might occur where the orientations of new evolution of the Central North Sea Rift. In: Fleet,
faults might be influenced by pre-existing basement A.J. & Boldy, S.A.R. (eds) Petroleum Geology of
fault patterns. Arguably, this could be the case in the Northwest Europe, Proceedings of the 5th Conference.
North Sea examples here, and, perhaps, is the reason Geological Society, London, 63–82, http://doi.org/
why their branch lines show a greater spread than in 10.1144/0050063
the West Cameron example. Færseth, R.B., Knudsen, B.E., Liljedahl, T. & Mid-
bue, P.S. & Suderstrum, B. 1997. Oblique rifting
This contribution was inspired by, and is dedicated to, Juan
and sequential faulting in the Jurassic development of
Watterson. Discussions with Dave Sanderson, Pete Bretan,
the northern North Sea. Journal of Structural Geology,
James Jackson and Dave Healy helped to focus the direc-
19, 1285–1302.
tion of the first draft of the manuscript. I am grateful to
Fossen, H. & Hesthammer, J. 1998. Structural geology
Atle Rotevatn, Richard Groshong and Bob Holdsworth
of the Gullfaks field, northern North Sea. In: Coward,
for their constructive comments on the manuscript,
M.P., Daltaban, T.S. & Johnson, H. (eds) Struc-
which led to an improved revised version. Stereograms
tural Geology in Reservoir Characterization. Geo-
were produced in Stereonet software (Cardozo & Allmen-
logical Society, London, Special Publications, 127,
dinger 2013).
231– 261, http://doi.org/10.1144/GSL.SP.1998.127.
01.16
References Fossen, H. & Rørnes, A. 1996. Properties of fault
populations in the Gullfaks Field, northern North
Aydin, A. & Reches, Z. 1982. Number and orientation of Sea. Journal of Structural Geology, 18, 179– 190.
fault sets in the field and in experiments. Geology, 10, Fraser, S.I., Robinson, A.M. et al. 2002. Upper Jurassic.
107–112. In: Evans, D., Graham, C., Armour, A. & Bathurst,
Badley, M.E., Price, J.D., Rambech Dahl, C. & Agdes- P. (eds) The Millennium Atlas: Petroleum Geology of
tein, T. 1988. The structural evolution of the northern the Central and Northern North Sea. Geological Soci-
Viking Graben and its bearing upon extensional modes ety, London, 157 –189.
of basin formation. Journal of the Geological Society, Groshong, R.H., Jr. 1988. Low-temperature deformation
London, 145, 455–472, http://doi.org/10.1144/gsjgs. mechanisms and their interpretation. Geological Soci-
145.3.0455 ety of America Bulletin, 100, 1329–1360.
Boyer, S.E. & Elliott, D. 1982. Thrust systems. Ameri- Healy, D., Jones, R.R. & Holdsworth, R.E. 2006.
can Association of Petroleum Geologists Bulletin, 66, Three-dimensional brittle shear fracturing by tensile
1196–1230. crack interaction. Nature, 439/5, 64–67.
Bretan, P., Yielding, G., Mathiassen, O.M. & Henza, A.A., Withjack, M.O. & Schlische, R.W. 2011.
Thorsnes, T. 2011. Fault-seal analysis for CO2 stor- How do the properties of a pre-existing normal-fault
age: an example from the Troll area, Norwegian Con- population influence fault development during a sub-
tinental Shelf. Petroleum Geoscience, 17, 181–192, sequent phase of extension? Journal of Structural
http://doi.org/10.1144/1354-079310-025 Geology, 33, 1312– 1324, http://doi.org/10.1016/
Brown, A.R., Edwards, G.S. & Howard, R.E. 1987. j.jsg.2011.06.010
Fault slicing – A new approach to the interpretation King, G. & Yielding, G. 1984. The evolution of a
of fault detail. Geophysics, 52, 1319– 1327. thrust fault system: processes of rupture initiation,
 Downloaded from http://sp.lyellcollection.org/ at University College Dublin on March 7, 2016

G. YIELDING

propagation and termination in the 1980 El Asnam Journal of Geophysical Research, 105, 25 553–
(Algeria) earthquake. Geophysical Journal of the 25 565.
Royal Astronomical Society, 77, 915–933. Nicol, A., Watterson, J., Walsh, J.J. & Childs, C.
Kneen, R.D. 2007. Interpret the fault network imaged in 1996. The shapes, major axis orientations and displace-
the West Cameron 3D seismic survey. MSc thesis, ment patterns of fault surfaces. Journal of Structural
Imperial College, London. Geology, 18, 235 –248.
Koestler, A.G., Milnes, A.G. & Storli, A. 1992. Oertel, G. 1965. The mechanism of faulting in clay
Complex hanging-wall deformation above an exten- experiments. Tectonophysics, 2, 343– 393.
sional detachment – example: Gullfaks Field, north- Reches, Z. 1978. Analysis of faulting in a three-dimen-
ern North Sea. In: Larsen, R.M., Brekke, H., sional strain field. Tectonophysics, 47, 109–129.
Larsen, B.T. & Talleraas, E. (eds) Structural Reymond, B. 1994. Three-dimensional sequence stratig-
and Tectonic Modelling and its Application to raphy offshore Louisiana, Gulf of Mexico (West Cam-
Petroleum Geology. Norwegian Petroleum Society eron 3D seismic data). Thèse de doctorat, Université
(NPF) Special Publication, 1. Elsevier, Amsterdam, de Lausanne.
243– 251. Roberts, A.M., Yielding, G. & Badley, M. 1990. A
Kostrov, V. 1974. Seismic moment and energy of earth- kinematic model for the orthogonal opening of the
quakes, and seismic flow of rock. Physics of the Solid late Jurassic North Sea rift system, Denmark–Mid
Earth, 1, 13–21. Norway. In: Blundell, D.J. & Gibbs, A.D. (eds)
Krantz, B., Ormand, C. & Freeman, B. 2014. Geo- Tectonic Evolution of the North Sea Rifts. Clarendon
logists get a closer ‘Look’ at complexity. AAPG Press, Oxford, 180– 199.
Explorer, February 2014, http://www.aapg.org/Pub Segall, P. & Pollard, D.D. 1980. Mechanics of discon-
lications/News/Explorer/Details/ArticleID/3788/ tinuous faults. Journal of Geophysical Research, 85,
Geologists-Get-a-Closer-‘Look’-at-Complexity 4337– 4350.
Krantz, R.W. 1988. Multiple fault sets and three- Walsh, J.J., Watterson, J., Bailey, W.R. & Childs, C.
dimensional strain: theory and application. Journal of 1999. Fault relays, bends and branch-lines. Journal of
Structural Geology, 10, 225–237. Structural Geology, 21, 1019–1026.
Kristensen, M.B., Childs, C.J. & Korstgård, J.A. 2008. Yielding, G., Badley, M.E. & Roberts, A.M. 1992.
The 3D geometry of small-scale relay zones between The structural evolution of the Brent Province. In:
normal faults in soft sediments. Journal of Structural Morton, A.C., Haszeldine, R.S., Giles, M.R. &
Geology, 30, 257– 272. Brown, S. (eds) Geology of the Brent Group. Geolog-
Maerten, L. 2000. Variation in slip on intersecting ical Society, London, Special Publications, 61, 27–43,
normal faults: implications for paleostress inversion. http://doi.org/10.1144/GSL.SP.1992.061.01.04