Interpretation of deep seismic data is challenging due to
the lack of direct geological constraints from drilling and the more limited
amount of data available from 2-D profiles in comparison to hydrocarbon
exploration surveys. Thus other constraints that can be derived from the
seismic data themselves can be of great value. Though the origin of most
deep seismic reflections remains ambiguous, an association between seismic
reflections and crustal strain, e.g. shear zones, underlies many
interpretations. Estimates of the 3-D orientation of reflectors may help
associate specific reflections, or regions of the crust, with geological
structures mapped at the surface whose orientation and tectonic history are
known. In the case of crooked 2-D onshore seismic lines, the orientation of
reflections can be estimated when the range of azimuths in a common midpoint
gather is greater than approximately 20
Deep seismic reflection surveys that image the entire continental crust are typically acquired as 2-D profiles due to cost and are able to provide subsurface images with a resolution of the order of 100 m or better. The interpretation of these deep seismic profiles, however, is often limited by the presence of reflections that can originate from locations out of the plane of the seismic profile, resulting in cross-cutting reflections in the migrated seismic section. In such situations it is difficult to identify which reflection, if any, should be included in an interpretation. Many onshore profiles have a crooked geometry because they are acquired along existing access roads. By using a 3-D travel time equation to determine the coherence of reflections, Bellefleur et al. (1997) showed how this limited 3-D geometry could be exploited to estimate the true 3-D orientation of subsurface reflectors where the acquisition line was particularly crooked, for example at sudden large bends in the road. Taking advantage of the increase in computing power over the last two decades, Calvert (2017) extended this method to every common depth point (CDP) in a crooked seismic profile, additionally providing quantitative estimates of the relative errors in the estimated angles of reflector dip and strike, and potentially also stacking velocity. These results, for example the angles and error estimates, are displayed as a function of time at each CDP on unmigrated seismic sections. Although it is possible to make general inferences on the distribution of subsurface reflectors, more detailed interpretation requires that the angle estimates be represented closer to their true subsurface position, i.e. on migrated seismic sections, which is an issue that was not addressed by Calvert (2017). The purpose of this paper is to present an approach to the migration of these reflector orientation attributes that allows their use in the interpretation of conventional 2-D migrated deep seismic sections; for example, by migrating more steeply dipping reflections into the middle crust, the predominant orientation of lower crustal reflections can be clarified. The importance of obtaining more accurate orientation estimates for positioning reflectors in 3-D, by for example deploying additional cross-line receivers, will also be discussed.
When a crooked seismic reflection line is processed, it is necessary to choose a slalom line through the distribution of source-receiver midpoints, and to define the CDP bins together with their dimensions along this line. Within a CDP bin, the conventional 2-D hyperbolic travel time equation may not accurately represent out-of-plane reflections due to the varying source-receiver azimuths. In these circumstances and under the straight-ray assumption used in stacking velocity analysis (Taner and Koehler, 1969), reflection travel times are better described by a 3-D travel time equation that includes the dip and strike of the reflector (Levin, 1971). When the seismic line is linear, the angles representing dip and strike cannot be uniquely determined, but along a crooked seismic profile, the distribution of source-receiver azimuths within a CDP gather varies, allowing the dip and strike to be well determined if a sufficiently large range of azimuths is present, for example where there is a large change in the direction of a seismic line (Bellefleur et al., 1997). In practice, most single CDP gathers on a crooked seismic line contain an insufficient number of traces, but this limitation can be mostly overcome by combining multiple CDP gathers into a much larger supergather that can be used for the estimation process; both Bellefleur et al. (1997) and Calvert (2017) provide examples of how the use of a large supergather permits the independent recovery of both dip and strike angles in many situations. The estimation method assumes that reflections within the supergather originate from a locally planar interface; as more CDP gathers are combined, this assumption can break down, especially where the geology is complex. For example where folded reflectors are present; for the crooked lines tested, supergathers of 40–80 CDPs appear to be adequate. If the algorithm were applied to every CDP gather with the output comprising the stacked trace computed using a moveout correction based on the estimated values of dip and strike, then this process could be viewed as an automated version of the cross-dip correction that is often applied manually to crooked seismic profiles, e.g. Nedimović and West (2003a) or Beckel and Juhlin (2018). It should, however, be noted that this cross-dip correction usually makes the assumption of linear moveout in the cross-dip direction within a CDP gather, which is not necessarily the case, especially where the line is particularly crooked.
Thus in a CDP gather and assuming a root mean square (RMS) velocity
function, the semblance of a reflection (Neidell and Taner, 1971) can be
calculated using a small time window, e.g. 40 ms, at each zero offset time
for a range of trial angles of dip and strike, which is measured from the north.
At each zero offset time, the estimated dip and strike correspond to the
angles with the maximum semblance, i.e. the most coherent reflection
(Bellefleur et al., 1997). Although the searched strike angle varies from
It is possible to apply 3-D prestack time migration to crooked 2-D reflection profiles; in some cases, for example where the deviation from 2-D is not great, the result is readily interpretable, but in others, the output 3-D volume can be dominated by artefacts from wave equation migration, with most structures incompletely imaged due to the limited amount of data recorded in the cross-line direction (Nedimović and West, 2003b). Thus crooked 2-D seismic profiles are usually migrated in 2-D for interpretation. To better integrate orientation angle estimates into the seismic interpretation it is therefore desirable to reposition these attributes in a way that is analogous to seismic data migration, so that they can be superimposed on their corresponding reflections. Attributes do not satisfy the assumptions necessary for wave-equation migration, and the result of applying such an algorithm to an unmigrated section containing attribute values would be meaningless. However, if the apparent dip of reflections on the unmigrated section is known, then a line migration or segment migration algorithm can be used to position the attribute value at a new output location corresponding to the migrated position of the corresponding reflection (Hagedoorn, 1954; Calvert, 2004). The sample value at each time and CDP location on the unmigrated section can be mapped to a small linear segment whose output position and dip is determined by the input position, apparent dip, and migration velocity (Raynaud, 1988). With seismic data, when multiple reflections are mapped to the same output location they are summed together, but for the attribute migration algorithm presented here, the output value is modified to be the attribute with the greatest semblance, implying that some less coherent attribute values will not be represented in the migrated output.
In principle, input attributes could be mapped to planar facets within a 3-D volume, but with narrow-azimuth crooked line surveys the uncertainties in determining the dip and strike along individual reflections are likely to be too large, resulting in the fragmentation of individual reflections after migration. Though 3-D migration is preferable in theory, the interpretation of an incomplete, sparse set of reflections in a 3-D volume is also likely to be challenging, and a better approach may be to forward model the appearance of 3-D structures in the crooked 2-D seismic profile.
This pragmatic approach to attribute migration is illustrated using a high-quality seismic line, 10GA-YU2, which was shot in 2010 over the Youanmi
Terrane of the Archean Yilgarn Craton of Western Australia as a
collaborative project between Geoscience Australia (GA) and the Geological Survey
of Western Australia (Fig. 1; Wyche et al., 2014). The Youanmi Terrane, which
contains several north-northeast striking greenstone belts, is the 3.05–2.70 Ga
core of the craton (Pigeon and Wilde, 1990; Van Kranendonk et al., 2013).
It is separated by the Ida Fault from the
Major terranes of the Yilgarn Craton in Australia with locations of deep seismic lines. The Youanmi Terrane represents the older core of the craton to which terranes of the Eastern Goldfields Superterrane were accreted during the late Archean. Greenstone belts are shown in green.
Line 10GA-YU2 was shot every 80 m using a source array of three Hemi60 vibrators and recorded by a 300-channel symmetric split spread with receiver groups every 40 m. A 12 s long Varisweep technique was used with either two or three sweeps recorded at each vibration point (VP). The seismic data were originally processed by Geoscience Australia using a conventional sequence of crooked-line geometry, refraction statics, geometric spreading correction, spectral equalization, velocity analysis, normal moveout, residuals statics, dip moveout correction, stretch mute, stack, and Kirchhoff migration; further details on the seismic acquisition and processing is provided by Costelloe and Jones (2014).
The preprocessing of the prestack seismic data for orientation analysis
included resampling to 8 ms, refraction statics, residual statics, amplitude
recovery with a
Line 10GA-YU2:
Values of local reflector strike can complement an interpretation based on a
conventional seismic section, and estimates of reflector strike along line
10GA-YU2 are shown in Fig. 2a, but only for reflections with a semblance
greater than 0.005 and for which the error in estimated strike angle is less
than 30
The 2-D migration of an orientation attribute requires two input datasets: the
attribute and an estimate of the apparent dip. To ensure consistency with
the conventional migration, the apparent dip was estimated from the
GA-processed stack section by determining the most coherent dip in a local
slant stack across an 800 m window at each time sample and CDP (Calvert,
2004). Using these apparent dips and the 1-D stacking velocity function,
each attribute sample was migrated to a 320 m long linear segment centred on
its output location with only the most coherent event retained at each
position, as described earlier. The length of the output segment was
selected to provide a degree of overlap for points migrated from the same
reflection, creating some continuity on the output image without producing
long linear segments that would be incapable of mimicking the geometry of a
curved reflector. The trace spacing of the input datasets was 40 m, and dips
greater than 50
The correspondence between reflections after frequency-wavenumber (F-K)
migration and the migrated strike attribute can be assessed by superimposing
the strike values on the migrated seismic data, as demonstrated by a section
at the east end of line 10GA-YU2 that shows the upper crust near the
boundary between the Youanmi Terrane and the Eastern Goldfields Superterrane
(Fig. 4). In general, local reflector strike estimates that appear laterally
consistent over
Section of line 10GA-YU2 across the boundary between the Youanmi
Terrane and the Eastern Goldfields Superterrane:
The estimates of reflector orientation are derived under the assumption that the reflector is locally planar. In the case of complex 3-D structures, for example a dome adjacent to the seismic line, the estimated strike will vary laterally, and inferring the nature of the subsurface structure will not be straightforward. In this situation, one approach would be to model in 3-D the seismic responses of a range of realistic features, estimate the local strikes from the synthetic data, and compare the synthetic results with the observations. However, where a large region of crust is dominated by reflectors with similar local strikes, then this characteristic and laterally extensive reflective fabric are likely to have arisen during a large-scale tectonic process. For example, a tectonic process such as crustal exhumation during shortening or the collapse of thickened crust can produce a thick band of pervasive seismic reflectivity that internally exhibits broadly similar reflector orientations. In this paper, we focus on the identification and interpretation of crustal domains in which a single reflector strike predominates.
Despite the absence of reliable estimates of reflector orientations at many
locations along line 10GA-YU2 due to the relatively straight road along
which the survey was carried out, it is possible to make some general
inferences on the distribution of reflector strike. Specifically, the
shallowly dipping reflections in the lower crust between 9 and 11 s are
commonly characterized by values that differ significantly from the
overlying middle crust; from CDP 6000–11 000 strikes range mostly from
80 to 110
The origin of the large amplitude reflection R1, which is important to the
interpretation summarized above, is unclear, because it truncates some
underlying reflections, but also appears to cut across others (Fig. 4).
After migration, estimates of reflector strike indicate that R1 exhibits a
strike of approximately 120
Along much of line 10GA-YU2, the range of source-receiver azimuths available
for the orientation analysis is quite limited, resulting in the exclusion of
many estimates due to their large errors. This problem is due to the mostly
linear geometry of the road along which the seismic line was shot. Deep
seismic lines are typically acquired along existing roads to minimize the
cost, which in the case of vibroseis surveys is often determined by the
number of shot points that can be acquired per day, i.e. the source effort.
When sufficient recording channels are taken to the field, the incremental
cost of deploying additional receivers can be relatively small. If
additional recording channels can be placed along crossing roads or readily
accessible land through which the survey passes, then the range of available
source receiver azimuths can be greatly increased, from
In this paper, a method of 2-D line migration that can be applied to any attribute continuously derived from seismic data has been presented. This algorithm uses the apparent dip obtained from the unmigrated stack section to move the attribute to a migrated position where it is represented by a short linear segment. (An alternative approach that iterates over an output 3-D migrated volume to identify the most coherent reflections would be much more costly and create artefacts, because an input location can contribute to multiple output locations.) Nevertheless the use of reflector orientation information to correctly position reflectors and their attributes throughout a 3-D volume, perhaps as planar facets, remains a long-term goal, but such an approach requires more accurate orientation estimates, which can be achieved by the use of additional off-line recording during 2-D onshore surveys.
By estimating and migrating the strike of subsurface reflectors along line 10GA-YU2, it has been possible to demonstrate that the lower crust of the eastern Younami Terrane of the Yilgarn Craton exhibits a systematic orientation of shallowly dipping reflectors which mostly dip to the N-NNE or S-SSW, in contrast to the middle crust which is characterized by a broad range of azimuths. Given that much of the crust here has been previously interpreted as reworked during extension and crustal collapse in the Late Archean, we suggest that the orientation of lower crustal reflections is consistent with approximately orogen-normal lower crustal flow at this time.
The seismic data are available from Geoscience Australia on request to clientrservcies@ga.gov.au.
AJC developed the seismic migration algorithms, interpreted the seismic data and wrote most of the paper; MPD interpreted the seismic data and provided the geological framework.
The authors declare that they have no conflict of interest.
This article is part of the special issue “Advances in seismic imaging across the scales”. It is a result of the 14th International Symposium on Deep Seismic Profiling of the Continents and their Margins, Cracow, Poland, 17–22 June 2018.
Tanya Fomin and Ross Costelloe prepared the SEGY files, which included the geometry and binning information necessary for this study. Ross Costelloe and Leoni Jones computed the static corrections during the initial processing that were applied prior to the orientation analysis. We thank Don White and Chris Juhlin for constructive reviews that improved the final paper. The field acquisition was funded by Geoscience Australia and the Geological Survey of Western Australia. This reprocessing project was supported by the Natural Sciences and Engineering Council of Canada. Michael P. Doublier is publishing with the permission of the CEO of Geoscience Australia.
This paper was edited by Michal Malinowski and reviewed by Don White and Christopher Juhlin.