Download Chapter 11 - Wide-angle refraction and reflection - U

In this chapter
11.1 Introduction
311
11.2 Wide-angle acquisition
311
Seismic sources 312
Receivers 314
11.3 Modelling wide-angle data
316
Travel-time analysis 317
Amplitude modelling 318
11.4 Examples of wide-angle seismic interpretations
Oceanic crust 319
Lithosphere stretching in the North Sea
Volcanic continental margins 321
Sub-basalt imaging 324
References
328
320
318
11
Wide-angle refraction
and reflection
R.S. White
Bullard Laboratories, Department of Earth Sciences, University of Cambridge,
Cambridge, United Kingdom
11.1 Introduction
Although near-offset seismic reflection profiling has dominated sub-surface
exploration and mapping for half a century, the wide-angle refractions (often
called ‘diving waves’) and reflections carry considerable additional information,
particularly on the seismic velocity of the sub-surface. This enhanced velocity
information may be crucial in understanding and imaging deep crustal structure.
Broadly speaking, conventional near-offset reflection data provide a good image
of the shapes of sub-surface interfaces, while the wide-angle data provide good
control on the velocities and hence lithologies of the sub-surface. Therefore,
normal incidence and wide-angle data provide complementary information.
Wide-angle data may also be of particular use for specialist problems such as
imaging sediments beneath high-velocity layers caused by basalt or tabular salt,
for undershooting complex structures such as those created by salt domes, and
for generating velocity models for pre-stack depth migration. In the following
sections, I discuss first the methods of acquiring wide-angle data, then the main
processing and modelling techniques used to aid interpretation of the data, and
finally show a variety of case examples that illustrate the power of the technique
in regional studies.
11.2 Wide-angle acquisition
The term ‘wide-angle’ is often used for arrivals at offsets greater than those
recorded by conventional seismic reflection surveys, capable of returning diving
waves (which are usually muted before conventional seismic reflection processing as they exhibit non-hyperbolic moveout), or wide-angle reflections
approaching or beyond the critical angle. The offsets required to record wideangle arrivals depend on the depth of the target: as a very crude rule-of-thumb,
Principles of Geologic Analysis DOI: 10.1016/B978-0-444-53042-4.00011-X
Copyright © 2012 by Elsevier B.V. All rights of reproduction in any form reserved.
311
Principles of Geologic Analysis
offsets of four to five times the target depth are necessary in order to record diving waves from the target depth. So, for example, in a whole-crustal study targeted at determining the Moho depth beneath a sedimentary basin, offsets of
over 100 km are required. If the target is relatively shallow, such as potentially
hydrocarbon-bearing sediments beneath basalt flows, offsets of 10 20 km
may suffice. But one common factor in all wide-angle profiles is that powerful
seismic sources are usually required, because seismic energy suffers a large
decrease in energy with offset due both to geometric spreading and absorption
along the travel path. Furthermore, as every cycle of a seismic wave suffers
absorption, the energy of high-frequency waves is reduced much more rapidly
than that of low-frequency waves in travelling the same distance. As a consequence, arrivals from deep in the crust or those that have travelled a long way
as diving waves tend to contain mainly low-frequency energy, as the high frequencies have been absorbed. For deep penetration, it is therefore necessary
to focus the seismic source to produce a waveform rich in low frequencies
(typically centred on about 10 Hz for deep crustal studies). A corollary of the
low-frequency nature of the seismic returns is that the achievable resolution is
of necessity reduced for deep or distant targets.
Seismic sources
The very earliest determinations of crustal structure used naturally occurring
earthquake sources with limited numbers of seismometers, usually in permanent
installations in observatories. Nevertheless, the major features of the crust and
upper mantle were identified using earthquake arrivals by the early twentieth
century. They included the definition of the Moho (strictly Mohorovočić discontinuity), now usually identified as the base of the crust, and the observation that
the continental crust is usually about 35 km thick, and considerably thicker
under mountain belts such as the Himalayas. Much of the nomenclature for seismic phases developed during these early studies remains in use today. Most
notably, PmP is used universally to denote a wide-angle reflection off the Moho,
Pg to denote a refraction through the crust (which was originally considered to
be of granitic composition, hence the subscript ‘g’) and Pn for the refraction
through the upper mantle beneath the Moho. Earthquake sources are large,
and the arrivals can be recognised over distances of hundreds of kilometres, so
they are well suited to whole-crustal regional studies. However, they are not so
useful for high-resolution studies of the upper crust or sedimentary basins, nor
are they useful for directed studies of particular regions. Therefore, for more
detailed studies, artificially generated seismic sources (often called ‘controlled
sources’) and portable seismometers were developed so that any desired region
could be targeted.
Interestingly, in recent years, the use of earthquakes as seismic sources has again
become popular, owing partly to the availability of relatively cheap but nonetheless high-fidelity three-component, broad-band seismometers, and partly to the
312
Principles of Geologic Analysis
increasing environmental regulation that makes it more difficult and expensive
to use large explosive charges, which are the natural controlled source replacement for earthquakes. Furthermore, earthquake sources are rich in shear waves,
whereas most controlled sources are poor at generating shear waves. The combination of information on compressional wave velocities and shear wave
velocities provides much better constraints on lithology than do compressional
wave velocities alone, and shear waves are particularly sensitive to the presence
of fluids, so are useful, for example, for studying hydrocarbon reservoirs.
Two main techniques that use earthquake sources are in common use today. The
first is to use microearthquakes to make 3D tomographic maps of small volumes.
This is only useful where the target area of interest is in the same region as the
earthquakes, which of course is rare: two settings where it does apply are
mapping regions around a sub-surface reservoir where there is active microseismicity as a result of hydrocarbon extraction, and around active volcanic regions.
The more useful method, which makes use of large global earthquakes at a
regional scale, is to map the main features of the crust and upper mantle using
arrays of broad-band seismometers by a technique known as mapping receiver
functions. Large earthquakes, typically with magnitudes greater than 6, are
required, and seismometers are usually left in place for 12 18 months or more
in order to record sufficient well-distributed global earthquakes with good signal
to noise ratios. The advantage of this method is that any chosen area can be
investigated as the earthquakes sources are global, and it is mainly just a matter
of waiting until a good suite of earthquakes has been recorded. The main limitation of this technique is that only low-frequency energy (1 Hz or lower) propagates globally, so the resolution is correspondingly limited. Other particular
circumstances that may limit the usefulness of receiver functions include the
presence of sedimentary sections that create multiples at the same time as the
primary arrivals of interest (often the Moho), and the absence of sharp interfaces
capable of generating the required mode conversions. The main use of receiver
functions is in mapping the depth of the Moho and of horizons within the upper
mantle.
For more detailed wide-angle acquisition on land, explosives remain the main
source capable of producing long-range seismic arrivals, and are still used for
regional studies. Swept-frequency Vibroseis causes somewhat less environmental
impact, but generally does not produce sufficient low-frequency energy to propagate more than a few tens of kilometers through the crust. So Vibroseis is most
useful for wide-angle studies of the uppermost crust, rather than the deeper or
whole-crustal section.
At sea, the use of airguns offers powerful, non-invasive seismic sources that can
produce waveforms tuned to specific targets by adjusting the chamber volume,
tow depth and firing sequence of individual guns in an array. For example, lowfrequency (c. 10 Hz) sources can be made by towing deep (typically 15 20 m),
313
Principles of Geologic Analysis
by using large chambers (up to 1000 cu. in. or even larger), or by tuning to
enhance the bubble pulse rather than tuning on the peak pulse as is conventional
(Lunnon et al., 2003). By using arrays of up to 50 guns totaling c. 10,000 cu. in.
volume, arrivals with good signal to noise ratio have been recorded to offsets of
more than 300 km (Staples et al., 1999). Further advantages of large airgun arrays
over explosive sources are the highly repeatable waveform available from airguns
and the ability to fire them at short intervals, thereby providing much denser shot
spacing than explosives.
Receivers
Because wide-angle arrivals usually require large offsets between the source and
the receiver, much wide-angle recording has used stand-alone, and relatively
sparse, receivers. At sea, following initial experiments with two ships, one of
which lowered a hydrophone and the other of which fired a shot, the technique
was developed of using free-floating sonobuoys with hydrophones suspended
beneath them (Fig. 11.1A). Typically 5 10 sonobuoys were deployed with
explosive or airgun shots fired out to any desired range. Of course, the sonobuoys usually drifted and data were sparse, so most interpretations using these
techniques assumed a 1D crustal model with no lateral variation, at least not
locally. The main improvement to this technique is to move the receivers to
the seabed, where they remain fixed (Fig. 11.1B). An added bonus of seabed
receivers is that three component seismometers can be recorded in addition to
a hydrophone, opening the possibility of recording shear (S) waves on horizontal
components in addition to compressional (P) waves and of removing water
multiples by PZ summation.
By the early twenty-first century, large deployments of up to 100 ocean bottom
seismometers (OBS) were common (e.g. White et al., 2002), in order to provide
better control on lateral inhomogeneity. For regional studies at sea, OBS remain
the receiver of choice. On land, the equivalent is free-standing 3-component
seismometers using a GPS signal to provide a reference time base.
The disadvantage of using free-standing seismometers is that of necessity the
recording array has to be relatively sparse and that it has a large spacing
between seismometers. This is adequate for determination of the velocity field,
but is not generally sufficiently dense for making a seismic image of the subsurface, such as is produced by conventional normal incidence or near-offset
reflection profiles. One partial solution to this problem is to use a seabed cable
(Fig. 11.1C). Such cables were initially only a few kilometres long, but with current technology it is possible to deploy up to 40 km in one length, with a surface
ship attached to record the multichannel data in real time. However, although
this allows true multichannel data to be recorded at wide angles, it is still
restricted to one location and it is difficult and expensive to move the
seabed cable.
314
Principles of Geologic Analysis
Figure 11.1
Methods of
acquiring wideangle seismic data:
(A) free-floating
sonobuoys with
suspended
hydrophones
(which may record
data internally, or
radio-telemeter it
to the firing ship);
(B) ocean bottom
seismometers,
usually with a
hydrophone plus
three-component
geophone, with
internal recording
and clock; (C) ocean
bottom cable, with
closely spaced
receivers (typically
25 m apart), both
hydrophone and
three-component
geophones, and
recording at a
surface seismic
vessel; (D)
expanding spread
configuration using
two ships to
produce a multifold
common mid-point
gather; (E) large
aperture acquisition
using multiple
passes of two
seismic vessels
sailing in line astern
at fixed separations
to synthesise large
offsets.
A Sonobuoys (hydrophone only)
B Ocean bottom seismometers (3 component + hydrophone)
C Ocean bottom cable (3 component + hydrophone)
D Two-ship expanding spread (common mid-point)
6000
PASS 1
PASS 2
6000
6000
6000
18000
6000
E Two-ship synthetic aperture profile (offsets 0–30,000 m)
The power of conventional streamer acquisition is the ease of obtaining multichannel coverage of the sub-surface. The development of digital, small-diameter
streamers has made it possible to tow ever-longer hydrophone streamers.
The increase from a maximum length of 2400 m in the 1970s to 12,000 m by
the end of the 1990s enables wide-angle data to be recorded using a single
ship with a source and towed streamer. With a 12,000 m streamer, critical angle
reflections may be recorded from the uppermost few kilometres below seabed,
which is adequate for many targets.
315
Principles of Geologic Analysis
However, two innovative strategies using two-ship acquisition have been developed that enable offsets to be recorded to any chosen range while still retaining
the advantages of a densely sampled multichannel streamer. The first of these is
called an ‘expanding spread configuration’ (Fig. 11.1D; Buhl et al., 1982). By sailing two seismic vessels away from a common mid-point on reciprocal courses,
the shots from one ship can be recorded across the streamer towed by the other
ship. After appropriate sorting and stacking, this produces essentially a giant
common mid-point gather out to any desired offset (limited only by the signal
to noise ratio). Such a profile is good for wide-angle interpretation, seamlessly
linking the normal incidence to the wide-angle profile. The main drawback of
an expanding spread profile (ESP) is that it provides poor control on lateral
variations in structure. Other than the shallow section, the assumption has to
be made of 1D (i.e., laterally uniform) velocity structure. Nevertheless, provided
care is taken to position the profile along a track with minimum lateral variability,
such as shooting along strike, this can be a useful way of rapidly acquiring multiple wide-angle profiles out to long offsets, each with high-density, stacked
data. A real example of such a survey comprising multiple ESPs is the determination of the crustal velocity structure across the continental margin west of
Rockall in the North Atlantic discussed in a later section.
The second technique is to steam two seismic vessels in line astern, two streamer
lengths apart, with the ships shooting alternately in a flip-flop fashion (Fig. 11.1E).
This is sometimes termed a ‘synthetic aperture’ profile as it is possible to
enhance the aperture of recorded data beyond that of a single streamer length.
Assuming reciprocity of travel times from source to receiver, the recorded multichannel data can be binned to produce a ‘supergather’ that mimics a single shot
into a streamer that is three times the length of either of the single streamers.
The apparent maximum offset of the supergather can be increased simply by
making a second and further passes of the two ships along the same profile at
successively greater separations (Fig. 11.1E). This technique retains all the
advantages of multichannel processing, although at the expense of recording
only half the fold of cover compared to a single ship acquisition, as a result of
the flip-flop shooting. An example of the use of this technique was the recording
of three passes of a pair of seismic vessels, giving offsets at all ranges up to
38,000 m on profiles over the Faroes shelf (White et al., 1999). These results
were targeted successfully at sub-basalt imaging, as discussed further in a later
section.
11.3 Modelling wide-angle data
The workhorse of standard multichannel seismic reflection processing is stacking
along trajectories with hyperbolic moveout. However, as angles of incidence
increase toward the critical angle, the simple hyperbolic assumption for reflections breaks down, even in the 1D case. The occurrence of anisotropy, which
316
Principles of Geologic Analysis
can reach 20% in some sediments, of velocity gradients and of laterally inhomogeneous structures, all cause the variation of travel time with offset to be much
more complicated than the simple hyperbolic case. The deviation from hyperbolic moveout becomes greater, the larger the offset and the farther the seismic
energy has travelled. The variations in travel times of diving waves are close to
linear with offset, although they too are equally subject to perturbations by
the effects of anisotropy, of velocity gradients and of lateral variations in structure. Therefore, different modelling strategies need to be applied to wide-angle
data than are generally used for conventional seismic reflection data.
Travel-time analysis
Most seismic modelling starts with matching the observed variation of travel
time against offset with the travel times predicted by a model. In the simplest
case, and in the early days of analysis before computers were sufficiently powerful, this was done using analytic solutions of the travel times, which almost
invariably assumed a 1D structure. As computers increased in power, and as
the recorded spatial data density increased, it became feasible to incorporate
into the models first 2D and more recently 3D variations in velocity structure.
Travel times through a trial model distribution of velocity are calculated using
a ray-tracer which uses Snell’s Law to calculate ray-paths through the model.
The model is then adjusted iteratively until there is a satisfactory fit between
the modelled and observed travel times. The model may either be adjusted
manually by an experienced operator, or an inverse method may be adopted,
which automatically modifies and iterates the model until the difference
between the observed and modelled travel times falls beneath a defined threshold. Normally, the iterations are continued until the travel times are fitted to
within their estimated picking uncertainty.
The bulk of ray-trace modelling is performed using only the compressional,
or P-waves. Exactly the same technique may be applied to shear waves, but
is far less commonly done, partly because explosive sources generate few
shear waves (and none if they are in water), and so it is necessary to rely on
the vagaries of mode conversion between P- and S-waves. Another reason that
shear waves are rarely used is that S-waves are slower than P-waves and so the
arrivals are always buried in the coda of earlier arriving energy and so are
harder to define and to pick. Nevertheless, shear waves are valuable in understanding the sub-surface because the ratio of P- to S-wave velocities of a rock is
far more diagnostic of the rock type and properties, and especially of the
presence of fluids such as hydrocarbons, than are the P-wave velocities alone.
Wide-angle data are far richer in converted S-waves than are near-offset
data, because mode conversion occurs most efficiently at oblique angles of
incidence, so S-waves are likely to become more important in the future with
the increasing availability of good three-component receivers (e.g., Eccles
et al., 2009).
317
Principles of Geologic Analysis
Travel times are
reduced at 7 km/s
(i.e., a linear
moveout of 7 km/s
has been applied to
the data). Note the
strong wide-angle
Moho reflection
and the marked
increase in
amplitude of this
reflection at about
40 km offset, as the
critical distance is
approached.
0
10
20
30
40
50
60
70
80
90
100
110
120
130
1
Time - Offset / 7 (s)
White et al., 2002).
Offset (km)
0
2
3
4
ve
wa
ter
Wa
Figure 11.2
Example of an
ocean bottom
wide-angle seismic
profile acquired
over the Norwegian
Sea and Faroes
continent-ocean
boundary (from
Crustal diving wave
flection
Mantle re
5
6
Water multiple
7
8
It is important to remember that the velocity model derived by travel-time
modelling is non-unique. In principle, there are numerous different models that
may fit the travel times equally well. Therefore, an important part of interpreting
wide-angle data is to investigate the extent of the model space that satisfactorily
fits the observations, and to explore the resolution of different areas of the model
by making extensive resolution tests. One way of doing this is to perturb the
travel times within the limits of their uncertainty, and then to make multiple
tomographic inversions. The average and variance of all these inversions give a
good measure of which features of the model are well constrained and which less
so: we show an example later (Korenaga et al., 2000; White and Smith, 2009).
Amplitude modelling
The wide-angle seismic energy carries far more information than the travel
times alone. Perhaps the most important characteristic is the amplitude of
the arrivals, and their variation with offset. As the critical angle is approached,
the amplitude increases greatly, as shown by the wide-angle Moho reflection
illustrated in Fig. 11.2. The exact form of the variation of amplitude with
angle of incidence is controlled primarily by the nature of the P- and S-wave
velocity contrasts across the interface causing the reflection. As with travel-time
modelling, use of this amplitude information started with 1D cases, progressing to more complex 2D and 3D models, and has developed from
forward modelling controlled by operator intervention to more rigorous inversion schemes.
11.4 Examples of wide-angle seismic
interpretations
In the following four examples, I show results from four typical and influential
wide-angle surveys that have had a high impact in geological understanding of the
structure of oceanic crust, of the development of rifted sedimentary basins, of continental margins, and of imaging through thick volcanic layers overlying sediments.
Each used different techniques which amplify the points discussed above.
318
Principles of Geologic Analysis
Oceanic crust
One of the major triumphs of wide-angle seismic profiling as early as the 1950s
was to show that oceanic crust is much thinner than continental crust by a
factor of 5 or more, that it varies relatively little in thickness across the globe
and that the characteristic velocities of the crust are quite unlike those found
in continental crust. All these observations were explained soon thereafter by
the theory of seafloor spreading. As oceanic crust covers 70% of the surface
of the earth, and wide-angle seismics were (and still are) the only way of measuring the crustal thickness, this was an early and crucial contribution of wideangle seismics to understanding the large-scale geology of the earth.
The classic work on oceanic structure was published first by Maurice Hill, based
at Cambridge in the United Kingdom. He used 41 wide-angle seismic profiles from the
Atlantic and Pacific oceans to define the mean crustal thickness (excluding the variable
sediment cover) as 6.4 1.9 km. By 1963, Russ Raitt, operating from Scripps Institution of Oceanography in the United States, had increased the number of profiles to
over 125, and the geographic coverage to include the Indian Ocean, and came up
with a similar average thickness of 6.6 1.6 km (Table 11.1). The acquisition methodology for these profiles was to use single sonobuoy receivers, sometimes with explosive sources, but more commonly with airguns, while the interpretation assumed a
small number of uniform velocity layers with plane interfaces between them.
It soon became apparent that the oceanic crust could be modelled by 3 layers,
with layer 1 the variable thickness sediments overlying layer 2, interpreted as
mainly extrusive basaltic volcanic rocks above a basal layer 3, later found to be
gabbroic igneous rocks. This prosaic division of the basement crustal section into
Table 11.1 Normal oceanic crustal structure from wide-angle seismic data
Raitt (1963) (using straight-line fits to travel time vs. distance)
Thickness (km)
Velocity (km/s)
Layer 2
1.71 0.75
5.07 0.63
Layer 3
4.86 1.42
6.69 0.26
8.13 0.24
Mantle
Igneous Crustal Thickness: 6.57 1.61 km
White et al. (1992) (using velocity structures constrained by amplitude modelling)
Layer 2
2.11 0.55
2.5–6.6
Layer 3
4.97 0.90
6.6–7.6
Mantle
>7.6
Igneous Crustal Thickness: 7.08 0.78 km
319
Principles of Geologic Analysis
layers 2 and 3 became standard nomenclature which has persisted for over 40
years, in all subsequent studies of oceanic crust.
The assumption of uniform velocity layers was a necessary consequence of the
simple analytical interpretation technique, which was to fit straight-line segments
to the first arrival diving waves on a distance versus travel-time plot. A new compilation of normal oceanic crustal structure by White et al. (1992), using results only
from well-controlled interpretations of profiles that had been modelled using
the amplitudes as well as the travel times of the wide-angle arrivals, yielded a surprisingly similar average thickness of 7.1 0.8 km. The increase of about 10% over
the thicknesses interpreted using simple straight-line fits to the travel-time data is
due to the amplitude-controlled interpretation constraining the velocity gradients
properly rather than assuming uniform velocity layers. Nevertheless, it is striking
that the first look at oceanic structure from wide-angle seismics in the midtwentieth century gave a surprisingly good picture of the structure of oceanic
crust, which underpinned the plate tectonic revolution and our understanding
of the role of mid-ocean spreading ridges in crustal formation.
Work over the past half-century has of course refined the understanding of oceanic
crustal structure and enabled us to map and to understand the causes of deviations
from the norm. The chief of these are the generation of thicker oceanic crust than
normal in the vicinity of thermal anomalies in the mantle caused by mantle
plumes, of locally thin crust in oceanic transform faults, sometimes associated with
serpentinisation, and of a reduction in crustal thickness as the full spreading rate
drops below a critical value of about 15 mm/a because of conductive cooling of
the upwelling mantle causing a reduction in the amount of melt generated.
Almost all this understanding has come from the use of wide-angle seismic profiles:
after 40 years of deep sea drilling, the deepest borehole into oceanic crust still only
reaches a little over 2 km below the seafloor, so our knowledge of the deeper in situ
section is still reliant primarily on seismic data.
Lithosphere stretching in the North Sea
In 1978, Dan McKenzie of Cambridge University, United Kingdom, published a
model for the development of sedimentary basins by lithospheric stretching and
thinning, which now underpins much basin research. This model was primarily
based on the observation that the thermal time constant of the subsidence which
provided accommodation space for the sediments in many basins was similar to
the thermal time constant for the subsidence of oceanic lithosphere, which is
formed by a well-understood process of rifting at mid-ocean ridges. Beneath sedimentary basins like the North Sea, there are large normal fault systems, showing
that extensional rifting occurred immediately prior to sediment infill. Therefore,
the available evidence pointed to an extensional cause for basin formation.
However, initially the model of lithospheric stretching was not universally
accepted, primarily because the normal faults imaged on seismic reflection profiles do not appear to supply sufficient extension for the observed subsidence
320
Principles of Geologic Analysis
(the solution to this apparent contradiction is that much of the extension is taken
up at a scale below that resolved by the seismic profiles, and that there is also
evidence of depth-dependent stretching, with the upper crust locally exhibiting
less extension than the whole lithosphere). The key to widespread acceptance of
the lithosphere stretching model was the proof from a wide-angle seismic profile
across the North Sea Basin that the crust was indeed thinned by extension under
the sedimentary basin, and that the amount and inferred timing of the thinning
was entirely consistent with the age, thickness and subsidence pattern of the
overlying sediments (Barton and Wood, 1984). As in the case of oceanic crustal
structure described above, this is an example of wide-angle seismics playing a
crucial role in the understanding and acceptance of a major step forward in
the way the earth is understood.
The wide-angle profile was recorded in 1981 1982 along a 530 km transect crossing the Central Graben of the North Sea, using 12 seafloor seismometers and 16
explosive shots ranging in size from 75 to 1000 kg (Fig. 11.3A). By modern standards, this is a small amount of data, but it demonstrated the key result of crustal
thinning under the centre of the basin (Fig. 11.3B). The travel times of the arrivals
were modelled using a ray-tracing technique through a 2D model, with further
constraints added from the major amplitude variations. The thinning is equivalent
to 70 km of stretching, which equates to a stretching factor (b) of 1.6. By the time
this profile was recorded, there was a good set of industry wells available, and
subsidence analysis of the drilled sedimentary sections confirmed the stretching
pattern mapped by the wide-angle seismic profile (Barton and Wood, 1984).
Volcanic continental margins
The third example of the impact of wide-angle seismics on advances in geological understanding of the planet comes from the rifted continental margins which
are produced between the oceanic and continental domains discussed in the
first two examples above. Rifted continental margins occur where the stretching
that generates a sedimentary basin does not stop, but continues until the
continental lithosphere is broken and a new oceanic basin forms between the
two portions of continental lithosphere. The examples shown here are from a
series of two-ship ESPs shot in 1985 across the Hatton continental margin west
of Britain on the northwest European continental margin, and a dip-line profile
shot using 85 OBS in 2002. The latter demonstrates the increase in velocity
resolution achieved using a denser OBS array.
The 1985 profiles are important because they were one of the first to show clearly
the presence of voluminous lower crustal intruded igneous rocks in addition
to large-scale extrusive volcanic flows on what are now known as ‘volcanic’ continental margins. The lower crustal igneous rocks are too deep to be drilled, and
never have been, but they were imaged clearly on the wide-angle seismic profiles
by their characteristic high (for continental lower crust) P-wave velocities
of 7.3 7.5 km/s. Such seismic velocities can be explained by the presence of
321
Principles of Geologic Analysis
59
N4
S7 S8
57
S6
F7
F4
(figures from Barton
and Wood, 1984).
N2
N5
C
Gr entr
ab al
en
S1
F2
Shots < 400 kg
Shots > 400 kg
PUSS array
55
A
2
0
4
2
6
crust
CE
N
TA
2.0
sediments
DIS
mantle
8.4
km / s
CI
TY
VE
LO
DEPTH
E
W
F2
5.8
6.2
20
40
F7
F4
0
km
Figure 11.3 Wideangle experiment
shot across the
North Sea Basin in
1981 1982 which
confirmed the
lithosphere
stretching model
for rifted basin
formation. (A) Map
showing position of
seabed pull-up
shallow-water
seismometers
(PUSSes), and
explosive shots
(stars), with main
normal faults
marked. (B) Final
velocity model
along profile with
velocity contour
interval 0.2 km/s,
sediments marked
by hatched lines,
no vertical
exaggeration
N1
N3
MOHO
S1
5.8
6.6
7.0
8.0
8.1
8.3
8.3
0
B
S6 S7 N1 S8 N2
8.1
8.3
N3
6.2
N4
N5
MOHO
6.6
7.0
50
DISTANCE
km
high-magnesium picritic melts generated by decompression melting of the
mantle by the interaction between rifting lithosphere and a thermal anomaly in
the mantle, often known as a ‘mantle plume’ or ‘hot-spot’ (White and Smith,
2009). The crucial factor that enables their presence to be identified in the deep
crust is their high seismic velocities, and these can only be determined with
wide-angle profiles.
The data acquisition in summer 1985 recorded seven ESPs orientated parallel to
the margin in order to reduce the effect of lateral variations in structure
(Fig. 11.4A). The receiver array was a 2400 m streamer with 25 m groups, and
322
Principles of Geologic Analysis
(figure adapted from
Fowler et al., 1989);
(C) velocity
structure from
Monte Carlo
average modelled
using tomographic
inversion of travel
times recorded by
OBS at positions
marked by ellipses
on the seafloor
(from White and Smith,
2009).
ICELAND
BASIN
60N
H
G
E
F
C
D
A
B
N
O
TT
A
H
59N
K
N
A
B
HATTON
BASIN
58N
km
0
22W
A
H
20W
50
F
E
OCEANIC
C
D
B
A
5.0
5.5
2.0
4.0
2.0
Depth (km)
16W
Location
G
6.5
10
100
18W
0
5.0
6.0
7.0
6.0
7.3
6.5
CONTINENTAL
COT
8.0
7.0
MANTLE
20
7.3
B Fowler et al. (1989)
8.0
0
5.5
6.0
5.0
4.5
5.0
Depth (km)
Figure 11.4
Crustal structure of
the Hatton-Rockall
volcanic continental
margin calculated
from a series of
two-ship expanding
spread profiles
(ESPs) shot along
strike over the
Hatton Bank
margin and
adjacent oceanic
crust, and from a
dense OBS profile:
(A) map showing
location of ESPs
(broken lines) with
dip profile down
their mid-points,
and location of OBS
profile (OBS at
circles); (B)
structure across the
continental margin
inferred from the
series of 1D
constraints on the
velocity-depth
variation provided
by the ESPs. Note
the lower crustal
igneous intrusions
that exhibit
characteristic
seismic velocities
above 7.3 km/s
10
COT
7.0
6.5
OCEANIC
MANTLE
7.5
7.4
CONTINENTAL
20
30
C White and Smith (2009) Monte Carlo average
NW
SE
100km
the profiles were shot twice using both explosives and airguns. The use of airguns
had the advantage of producing a conventional seismic reflection profile at the
same time as the ESPs were recorded, while the explosive shots ensured penetration to the base of the crust. With modern larger and much more powerful airguns
capable of producing arrivals that can be recorded to ranges of over 120 km (e.g.,
Fig. 11.2), the use of explosives is no longer necessary for such targets. The second,
OBS profile (Fig. 11.4C) used a 6340 cubic inch (104 l) array of airguns.
A cross-section of the structure across the rifted margin (Fig. 11.4B) shows the
main results from modelling two-ship ESPs (Fowler et al., 1989). The nearby
parallel OBS profile (Fig. 11.4C) shows the average from 100 automated inversions, with small perturbations within the estimated picking uncertainty applied
to the travel-time data prior to each inversion (White and Smith, 2009). The final
velocity model is therefore a robust measure of the structure and shows better
resolution of the high-velocity region in the continent-ocean transition. The
323
Principles of Geologic Analysis
crust thins from the continental to the oceanic end, and extrusive volcanics in
the upper section just beneath the seafloor are imaged by the seismic reflection
profiles and by their P-wave velocities of 4.0 5.5 km/s, while the multiple sill
intruded lower crust is demarcated by the region of high seismic velocities above
7.3 km/s. Similar results have been found subsequently on both the Greenland
margin that is conjugate to the Hatton margin, and on many other volcanic
margins worldwide, including the northern North Atlantic, the South Atlantic
and the Indian continental margins. In all these cases, wide-angle seismic profiles
continue to provide the best constraints on the volume of igneous rock added to
the crust, and provide a crucial constraint in modelling the development and
subsidence of the continental margins and their hinterland.
The improved velocity field from wide-angle seismic profiles allows better
pre-stack depth migrations to be made of conventional seismic reflection profiles.
I show an application of this in Fig. 11.5, which is across the North Atlantic volcanic continental margin near the Faroe Islands (White et al., 2008). The velocity
structure across the continent-ocean transition (Fig. 11.5A) is remarkably similar
to that found on the Hatton Bank margin (Fig. 11.4C), with abnormally high
velocities in the lower crust. However, it also shows basalt flows (in green,
Fig. 11.5A) extruded above sediments (in blue, Fig. 11.5A) beneath the Fugloy
Ridge. On the pre-stack depth migrated reflection image (Fig. 11.5B), the
extruded lavas produce seaward-dipping reflectors in the upper crustal section,
and the intrusions produce sub-horizontal layering visible in the mid-crust.
Sub-basalt imaging
The previous case examples were from, respectively, oceanic crust, continental
crust and the boundary between the two found at continental margins. In addition to illustrating the importance of wide-angle data in understanding the geology of three of the earth’s major settings, they also demonstrate the use of three
of the main techniques for seismic acquisition. The oceanic crust studies used
primarily sonobuoys (Fig. 11.1A), and the North Sea continental basin study
used seabed receivers (Fig. 11.1B) while the volcanic continental margin profiles
used two-ship expanding spread and dense OBS profiles (Fig. 11.1B and D).
The final example of the use of wide-angle data discussed in this section used
the two-ship method of recording synthetic aperture profiles (Fig. 11.1E), in this
case with three passes giving offsets from 0 to 38,000 m. The target for this was
primarily the potentially hydrocarbon-bearing sub-basalt sediments. This example illustrates well the integration between near-normal incidence profiling
and wide-angle profiling, as data from the near and far offsets are combined
seamlessly to improve both the velocity resolution (which carries information primarily on rock types), and the reflection images (which carry information
primarily on the geometric structure of the sub-surface). Combination of the
near-offset and wide-angle data gives much better understanding of the subsurface than using either alone, and each benefits from the other.
324
Norwegian
Basin
0
Fugloy
Ridge
Faroe-Shetland
Basin
Depth (km)
5
10
15
20
mantle
mantle
25
NW
A
SE
30
0
50
100
150
200
250
0
Two-Way Time (s)
seaward-dipping
reflectors
lower-crustal
intrusions
5
10 NW
SE
0
50
100
150
Distance from Chron 22 (km)
B
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
200
6.0
6.5
250
7.0
7.5
Velocity (km/s)
Figure 11.5 (A) Seismic velocity structure of crust across the continent-ocean transition near the Faroe Islands.
Velocities of post-rift Cenozoic sediments (purple) are constrained by semblance moveout analysis of reflections
recorded on the 12-km-long hydrophone streamer and sub-sediment crustal velocities are constrained by
tomography using wide-angle reflections and diving waves from ocean bottom seismometer data (OBS locations
shown by circles at seafloor). Colour bands are at 0.1 km/s intervals, with contours above 7.0 km/s spaced every
0.1 km/s to highlight the lower-crustal velocity. (B) Section of multichannel seismic reflection profile with
superimposed P-wave velocity field from (A). Extrusive lava flows are imaged as seaward-dipping reflectors
between 50–90 km along profile and as subhorizontal layered basalts at distances >90 km. Lower-crustal
layering coincident with high (>7.0 km/s) velocities, caused by igneous intrusions, lies beneath the seaward-dipping
basalts on the continent-ocean transition, with a termination at c. 90 km against continental crust with lower
velocities. Processing of the reflection profile included source designature, multiple suppression and post-stack
time migration (diagram modified from White et al., 2008).
325
7W
6W
5W
4W
3W
2W
62N
1
7
61N
500
A
60N
6W
0
Depth (km)
0
10
5W
4W
3W
40
Distance (km)
80
4.80
basalt
5.20
5.50
5.65
6.00
2W
120
2.20
3.80 4.20
5.80
160
sed.
basement
6.78
20
7.00
6.80
7.80
mantle
30
7.20
7.80
FLARE-1
40
B
Faroe Is.
Depth (km)
0
Faroe Shelf
Faroe-Shetland Basin
basalt
sed.
10
basement
20
30
C 40
intruded?
mantle
FLARE-1
Figure 11.6 (A) Map showing location of long-offset two-ship profiles 1 and 7 of the FLARE project. Purple line
marks southeastern limit of Tertiary basalt flows. Bathymetric contours every 100 m, with heavier contour at
1000 m. Inset shows location of survey area on the Northwest European continental margin; black lines mark
location of all 12 FLARE profiles. (B) Representative wide-angle ray-paths through the crustal velocity model along
the FLARE-1 profile, including both offshore supergathers and onshore land station recording on Suduroy. Red rays
are those turned in the basalt layers, or reflected off their bottom. Blue rays are reflected off the top of the basement
which marks the base of the sub-basalt sediments, or returned as diving rays through the top of the basement.
Green rays are reflections off the Moho. Only every tenth ray is shown for clarity. Selected velocities shown in km/s.
(C) Interpreted cross-section. Interfaces constrained well by seismic data are shown as solid lines. Note the crustal
thinning eastward toward the centre of the Faroe-Shetland Basin, and the basalt flows (red) extending eastward
from their source in the vicinity of the Faroe Islands and over-riding a pre-existing sedimentary section (yellow)
which lies above the basement (brown). Green section beneath Faroe Shelf and Islands represents basement with
velocities modified by weathering, igneous intrusion or emplacement of tuffs.
326
Principles of Geologic Analysis
The Faroes Large Aperture Research Experiment (FLARE) profiles were recorded
in 1996 by an industry consortium (White et al., 1999), across a transect of the
Faroes Shelf with large sub-surface basalt flows extending southeastward from
the Faroes Islands (Fig. 11.6A). Travel-time tomography using a ray-tracing
method showed unambiguously, for the first time in this area, that a thick sediment layer lay beneath the basalts, and that the basalt layer thinned from several
kilometres thick near the Faroe Islands to a feather edge some 150 km away
(Fig. 11.6B). The 38 km maximum offsets provided by the synthetic aperture profiles were sufficient to provide good wide-angle arrivals from the basalts, and from
the sub-basalt basement, but did not reach sufficiently great offsets to give strong
wide-angle arrivals from the Moho at the base of the crust. However, on the profile
illustrated here, fixed three-component land seismometers were deployed at one
end of the profile, the Faroese Island of Suduroy, which provided strong arrivals to
much greater offsets (Richardson et al., 1999). These constrained the Moho depth
over a short segment of the profile, which then was extrapolated along the full
length of the profile by using measurements of the gravity anomaly (Fig. 11.6C).
Measurement of the velocity structure provides powerful insights into the geology of this region, in this case showing the presence of sub-basalt sediments.
However, the good velocity control provided by the wide-angle data also
0
SE
NW
FL A RE-1
1
Velocity (km/s)
Depth (km)
2
Basalt
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
3
4
Sediment
5
6
7
Basement
8
9
0
0
WSW
0
20
40
80
100
20
Distance (km)
40
60
80
60
100
ENE
120
FL A RE-7
1
2
Depth (km)
Figure 11.7
Composite profiles
(from White et al.,
2003) along FLARE-1
(down dip) and
FLARE-7 (along
strike; see Fig. 11.6A
for location),
generated by
merging the
conventional depthconverted nearoffset seismic section
above the base of
basalt with the
pre-stack depth
migrated base-basalt
and sub-basalt
arrivals, and the
pre-stack depthmigrated basement
arrivals. Colours
corresponding to
velocities derived
from ray-tracing
modelling and used
for pre-stack depth
migration are
superimposed on the
seismic profiles.
3
4
5
Basalt
Sediment
6
7
8
Basement
9
327
Principles of Geologic Analysis
provides two other major benefits. First, it enables a well-controlled pre-stack
depth migration of the conventional seismic data to be made, markedly improving the image. Second, the wide-angle arrivals have large amplitudes because
they approach the critical angle, so they themselves can be migrated back to
normal incidence to provide good images, such as those shown in Fig. 11.7,
of reflection horizons (in this case the base of the basalt layer and the underlying
basement) that are too weak to image well with conventional near-offset seismic
data (White et al., 2003).
References
Barton, P.J., Wood, R.J., 1984. Tectonic evolution of the North Sea basin: crustal stretching and
subsidence. Geophys. J. R. Astron. Soc. 79, 987 1022.
Buhl, P., Diebold, J.B., Stoffa, P.L., 1982. Array length magnification through the use of multiple
sources and receiving arrays. Geophysics 47, 311–315.
Eccles, J.D., White, R.S., Christie, P.A.F., 2009. Identification and inversion of converted shear
waves: case studies from the European North Atlantic continental margins. Geophys. J.
Int. 179, 381–400. doi:10.1111/j.1365-246X.2009.04290.x.
Fowler, S.R., White, R.S., Spence, G.D., Westbrook, G.K., 1989. The Hatton Bank continental
margin-II. Deep structure from two-ship expanding spread seismic profiles. Geophys. J.
96, 295–309.
Korenaga, J., Holbrook, W.S., Kent, G.M., Kelemen, P.B., Detrick, R.S., Hopper, J.R.,
Dahl-Jensen, T., 2000. Crustal Structure of the Southeast Greenland margin from joint
refraction and reflection seismic tomography. J. Geophys. Res. 105, 21591–21614.
Lunnon, Z.C., Christie, P.A.F., White, R.S., 2003. An evaluation of peak and bubble tuning in
sub-basalt seismology: modelling and results. First Break 21, 51–56.
Raitt, R.W., 1963. The crustal rocks. In: Hill, M.N. (Eds.), The Sea, vol. 3. Wiley Interscience, New
York, pp. 85 102.
Richardson, K.R., White, R.S., England, R.W., Fruehn, J., 1999. Crustal structure east of the Faroe
Islands. Pet. Geosci. 5, 161–172.
Staples, R.K., Hobbs, R.W., White, R.S., 1999. A comparison between airguns and explosives as
wide-angle seismic sources. Geophys. Prospect. 47, 313–339.
White, R.S., McKenzie, D., O’Nions, R.K., 1992. Oceanic crustal thickness from seismic measurements and rare earth element inversions. J. Geophys. Res. 97, 19683–19715.
White, R.S., Fruehn, J., Richardson, K.R., Cullen, E., Kirk, W., Smallwood, J.R., Latkiewicz, C.,
1999. Faroes Large Aperture Research Experiment (FLARE): imaging through basalts. In:
Fleet, A.J., Boldy, S.A.R. (Eds.), Petroleum Geology of Northwest Europe: Proceedings of
the 5th Conference. Geological Society of London, London, pp. 1243–1252.
White, R.S., Christie, P.A.F., Kusznir, N.J., Roberts, A., Davies, A., Hurst, N., Lunnon, Z.,
Parkin, C.J., Roberts, A.W., Smith, L.K., Spitzer, R., Surendra, A., Tymms, V., 2002. iSIMM
pushes frontiers of marine seismic acquisition. First Break 20, 782–786.
White, R.S., Smallwood, J.R., Fliedner, M.M., Boslaugh, B., Maresh, J., Fruehn, J., 2003. Imaging
and regional distribution of basalt flows in the Faroe-Shetland Basin. Geophys. Prospect. 51,
215–231.
White, R.S., Smith, L.K., Roberts, A.W., Christie, P.A.F., Kusznir, N.J., iSIMM Team, 2008. Lowercrustal intrusion on the North Atlantic continental margin. Nature 452, 460–464.
doi:10.1038/nature06687.
White, R.S., Smith, L.K., 2009. Crustal structure of the Hatton and the conjugate east
Greenland rifted volcanic continental margins, NE Atlantic. J. Geophys. Res. 114, B02305,
doi:10.1029/2008JB005856.
328