Geophysical Prospecting, 2006, 54, 709719

Section I Madrid Workshop on Near-Surface 2005

CRS strategies for solving severe static and imaging issues in seismic
data from Saudi Arabia
G. Gierse, J. Pruessmann and R. Coman
TEEC, Burgwedeler Strasse 89, D-30916 Isernhagen, Germany

Received January 2006, revision accepted April 2006

ABSTRACT
Static shifts from near-surface inhomogeneities very often represent the key problem
in the processing of seismic data from arid regions. In this case study, the deep bottom
fill of a wadi strongly degrades the image quality of a 2D seismic data set. The resulting
static and dynamic problems are solved by both conventional and common-reflection-
surface (CRS) processing. A straightforward approach derives conventional refraction
statics from picked first breaks and then goes through several iterations of manual ve-
locity picking and residual statics calculation. The surface-induced static and dynamic
inhomogeneities, however, are not completely solved by these conventional methods.
In CRS processing, the local adaptation of the CRS stacking parameters results in
very detailed dynamic corrections. They resolve the local inhomogeneities that were
not detected by manual picking of stacking velocities and largely compensate for
the surface-induced deterioration in the stack. The subsequent CRS residual statics
calculations benefit greatly from the large CRS stacking fold which increases the
numbers of estimates for single static shifts. This improves the surface-consistent
averaging of static shifts and the convergence of the static solution which removes
the remaining static shifts in the 2D seismic data. The large CRS stacking fold also
increases the signal-to-noise ratio in the final CRS stack.

velocity variation cannot be obtained from normal-moveout

INTRODUCTION
Strong variations in both near-surface lithology and topogra-
phy cause problems in seismic data processing in many parts of
the world. These difficulties culminate in arid regions where
weathering layers with extremely large thicknesses and low
velocities alternate with outcrops of consolidated rock at the
surface. The resulting strong variations in seismic traveltimes
may totally obscure the subsurface reflections, thus emphasiz-
ing the solution of the static problem as the key step in seismic
processing.
In general, the thickness of the weathering layer is not re-
solved by the reflection seismic data and thus the near-surface
E-mail: info@teec.de
(NMO) analysis. Hence, it is common practice to obtain the
velocity of the weathering layer from uphole measurements.
Using these starting values, a layered model of the near surface
is then derived from refracted first arrivals. First-arrival times
are available either directly from the reflection seismic data or
from special near-surface refraction measurements.
A 2 D S E I S M I C D ATA S E T F O R T E S T I N G
S TAT I C M E T H O D S
There is an urgent need for an improvement in the costly and
often insufficient derivation of static corrections, especially
since a major portion of the worlds oil and gas reserves are
located in arid regions. This applies particularly to the huge


C 2006 European Association of Geoscientists & Engineers 709
710 G. Gierse, J. Pruessmann and R. Coman
Figure 1 Stack of complete test line with elevation statics applied (horizontal length: 42.6 km). Only the section of length 14 km on the left was
selected for processing in this case study.
arid belt extending from North Africa to the deserts of the Ara-
bian Peninsula. In their efforts to find better solutions for the
static problem, Saudi Aramco released a 2D seismic data set
from Saudi Arabia, in order to focus the activities of research
groups world-wide on a specific static problem. The results of
various groups and approaches, including this work by TEEC,
were presented in a special workshop at the 2005 EAGE meet-
ing in Madrid.
The 2D seismic data set was recorded by 240 channels lo-
cated at intervals of 30 m in a symmetric split-spread ge-
ometry without a gap. Dense off-line shooting with vibro-
seis sources led to a high subsurface fold of 120 and to off-
sets in the range of approximately 50 m to 3000 m. The
seismic line was acquired over rough topography with eleva-
tions varying between 460 m and 600 m. Figure 1 shows the
variability of the elevation together with a preliminary stack

C 2006 European Association of Geoscientists & Engineers, Geophysical Prospecting, 54, 709719
for the complete test line. Ridges of consolidated rocks alter-
nated with steep valleys of wadis that contained deep bottom
fills of debris, indicating strong variations in the near-surface
velocity.
In this case study, only the northern part of the line with a
length of about 14 km was processed, as indicated in Fig. 1.
This part was selected since it covered both a well-resolved
region below the ridge, which served as a reference in the
processing, and a strongly disturbed region below the wadi
where the compensation for near-surface effects by common-
reflection-surface (CRS) strategies could be tested. The se-
lected part is illustrated in the zoom section of the preliminary
stack in Fig. 2.
On the selected part of the line, the strong variations in the
near-surface velocity can be observed in the raw shot data of
Fig. 3. The picked velocities of the first refractor varied in
 CRS strategies for solving severe static and imaging issues 711

Figure 2 Stacked section with elevation statics applied (zoom from Figure 1, length 14 km).

Figure 3 Raw shots of the 2D data set with strong variations in first-break velocities and data quality.


C 2006 European Association of Geoscientists & Engineers, Geophysical Prospecting, 54, 709719
712 G. Gierse, J. Pruessmann and R. Coman
the range of approximately 2000 m/s to 4000 m/s. Moreover,
the effect of the bottom fills on refracted first arrivals can be
clearly seen in Fig. 3, as well as the strong variations in data
quality.
The main problem for processing the data was thus clearly
presented by these effects of the near surface. The solution
of this problem, however, was even more difficult, since there
was no information available on the near-surface layer. As
a typical example of vibroseis data, there were no uphole
times available from shooting, and neither supplementary
measurements nor any other information were supplied. In
addition, the near-offset traces at about 50 m did not allow
velocity estimation for the very near surface from first-break
picks.

C 2006 European Association of Geoscientists & Engineers, Geophysical Prospecting, 54, 709719
Figure 4 Near-surface velocity profile from
first-break picks.
Figure 5 Source statics calculation in a one-
layer refraction model from picked first
breaks.
Hence, this paper shows a rather straightforward approach
using conventional statics methods in the first phase and the
CRS technique in a second phase. In the conventional ap-
proach, basic statics were derived from refracted arrivals ob-
tained by first-break picking, followed by standard residual
statics calculation and velocity analysis. In the second phase,
the results were refined by CRS residual statics combined with
CRS velocity optimization.
CONVENTIONAL PROCESSING
A N D S TAT I C S O L U T I O N
The processing was based on the raw shots and the geometry
information, which were supplied by Saudi Aramco for the
 CRS strategies for solving severe static and imaging issues 713

Figure 6 CMP stack after conventional statics processing (horizontal length: 14 km).

2D seismic line. The work was focused on the northern part
of the line between CDP 3081 and CDP 4000. The initial
signal processing included a Radon removal of steep dipping
noise, compensation for spherical divergence, and single-trace
spike deconvolution. Figure 2 shows the stacked section for
the selected part of the line with only elevation statics applied.
The data was corrected to a flat datum at 500 m. The elevation
profile along the line is given in the top of the figure showing
the rough topography. It is obvious that the bottom fill in the
wadi on the right-hand side of the section completely obscures
the image of the subsurface.
In order to estimate the amount of near-surface veloc-
ity variation, an initial velocity model was calculated from
first-break picks and is shown in Fig. 4. The velocity de-
creases significantly below the valley on the right of the
section.
The first-break times obtained by manual picking were en-
tered into an industry-standard refraction statics tool. This
refraction statics solution involved the construction of a near-
surface velocitydepth model from refracted arrivals, which
was then used to calculate the long-wavelength trend for the
basic statics.
The corresponding intermediate results and the final refrac-
tion statics solution for the sources are shown in Fig. 5. The
elevation (in m) of Fig. 5(a) was converted into the eleva-
tion statics (in ms) of Fig. 5(b) using a replacement velocity of
3500 m/s. As expected, the refraction statics (in ms) in Fig. 5(b)
resemble the elevation statics at the ridge with outcrops of the
bedrock on the left, but there is strong deviation in the wadi on
the right with its bottom fills of debris. Figure 5(c) shows the
depth to the consolidated refractor (in m), which is small on
the ridge, but large in the wadi. The refractor velocity (in m/s)
is shown in Fig. 5(d). For the one-layer case that was assumed
in the refraction statics calculation, the refractor velocity in
this part of the line varies from about 33004500 m/s. For
the weathering layer above the refractor, a constant velocity
of 1000 m/s was assumed, since no information was available
here.


C 2006 European Association of Geoscientists & Engineers, Geophysical Prospecting, 54, 709719
714 G. Gierse, J. Pruessmann and R. Coman
Figure 7 CMP stack with stacking velocity from CRS parameters (horizontal length: 14 km).
After establishing the low-frequency statics from refracted
first arrivals, the remaining high-frequency time shifts had to
be removed by residual statics. In the conventional approach,
they were obtained by a correlation technique similar to the
residual statics method of Ronen and Claerbout (1985). The
residual statics were improved in several iterations, together
with manual stacking velocity analysis.
The refraction statics and residual statics from this conven-
tional processing finally produced the stack shown in Fig. 6.
Comparison with the stack in Fig. 2 obtained from eleva-
tion statics shows a good restoration of the flat reflections
below the wadi on the right. Obvious static problems remain
in the centre of Fig. 6, where the steep slope down to the
wadi prevented shooting. Additionally, a change in charac-
ter is observed especially at deeper reflected events at 800
1600 ms. Below the wadi, the signal frequency and continuity
of the events decreases significantly in comparison with the
less disturbed seismic data recorded on consolidated ground
on the left.

C 2006 European Association of Geoscientists & Engineers, Geophysical Prospecting, 54, 709719
CRS METHOD
The common-reflection-surface (CRS) method is a macro-
model independent imaging method in the time domain.
Refining similar approaches by de Bazelaire (1988) and
Gelchinsky (1988), it was developed by Hubral (1999a,b)
and his group at the University of Karlsruhe. Success-
ful applications in exploration projects were reported by
Trappe, Gierse and Pruessmann (2001) and Gierse et al.
(2003).
The CRS stacking parameters are automatically determined
for each sample of the zero-offset stack section. The optimum
parameter values are characterized by a semblance maximum
along the associated stacking surface. This optimization com-
prises the imaging of dipping structures since dip is part of the
subsurface model assumptions.
Manual picking of stacking velocities is, in principle, not
required but is helpful to guide the automatic determina-
tion of the CRS stacking parameters. Similarly, the stacking
 CRS strategies for solving severe static and imaging issues 715

parameters deliver a stacking velocity field that is locally
adapted to the data.
Both the partial automation of the imaging process and
the local optimization of the imaging result are advantages of
CRS stacking in comparison with the conventional common-
midpoint (CMP) stacking technique. The CMP stacking hy-
perbolae assume reflections from an idealized subsurface with
plane horizontal horizons. The CRS stacking surfaces, in
contrast, take into account arbitrary dips and curvatures
of the subsurface horizons and thus represent a much bet-
ter approximation to the traveltime surfaces of the reflected
wavefield. Moreover, the CRS stacking surface stacks the
reflection from a subsurface element across several neigh-
bouring CMP locations, whereas CMP stacking is con-
fined to single CMPs. This implies both a better imag-
ing of complicated subsurface structures including strong
dips, and an increased signal-to-noise ratio of the stacked
signal.
A detailed description of the CRS technique is given by Jager
et al. (2001) and Mann (2002).

Figure 8 Stacking velocity fields: (a) from

manual velocity analysis (b) from CRS pro-
cessing.


C 2006 European Association of Geoscientists & Engineers, Geophysical Prospecting, 54, 709719
716 G. Gierse, J. Pruessmann and R. Coman
CRS PROCESSING
The CRS processing of the 2D seismic data from Saudi Arabia
was based on the conventional processing described above. It
started from the static and dynamic corrections obtained by
conventional methods. The static corrections and mute func-
tions were applied to the CMP-sorted prestack data before
entering the CRS processing. Similarly, the stacking velocity
from conventional velocity analysis was provided as a guide
function for the CRS stacking parameter search.
The local determination of the CRS stacking parameters
provided a detailed velocity fluctuation around the input
stacking velocity from conventional processing. A CMP stack
based on these CRS stacking velocities is shown in Fig. 7.
Comparison with the conventional CMP stack of Fig. 6
shows a distinct improvement in reflected events below the
wadi on the right side of the section. The decrease in signal
frequency and event continuity that was observed in Fig. 6 has
now almost disappeared.
Figure 9 CMP stack with final CRS residual statics (horizontal length: 14 km).

C 2006 European Association of Geoscientists & Engineers, Geophysical Prospecting, 54, 709719
The change in signal character thus turned out to be a veloc-
ity effect. Figures 8(a.b) compares the stacking velocity fields
used in Figs 6 and 7, respectively. Obviously, the heteroge-
neous surface conditions in the wadi led to local distortions
of the stacking velocity at greater depths. The smooth velocity
field obtained from manual velocity analysis in Fig. 8(a) failed
to incorporate these distortions on the right of the section but
was sufficient at the less complicated consolidated surface on
the left of the section. The locally adapted velocity field ob-
tained from CRS processing analysis in Fig. 8(b), however, pro-
vides an excellent determination of these inhomogeneities. In
general, the CRS velocity field is more event-consistent, lead-
ing to increased horizontal continuity but decreased vertical
smoothness.
Further improvements in the stack can be obtained from the
large CRS stacking fold which is due to large stacking aper-
tures across several CMP locations. In order to avoid smearing
effects, the aperture may be limited to the Fresnel zone which
can be estimated from the CRS stacking parameters. The first
 CRS strategies for solving severe static and imaging issues 717

Figure 10 Detailed effect of CRS residual statics by comparison of CMP stacks before application (top: zoom from Figure 7) and after application
(bottom: zoom from Figure 9).


C 2006 European Association of Geoscientists & Engineers, Geophysical Prospecting, 54, 709719
718 G. Gierse, J. Pruessmann and R. Coman
Figure 11 Final CRS stack (horizontal length: 14 km).
Fresnel zone comprises the area of a reflector that scatters
constructively interfering portions of energy back to a surface
point. The Fresnel zone increases both with traveltime and
with wavelength.
The advantages of the increased fold are first exploited by
the CRS residual statics approach. Each prestack trace con-
tributes to several image traces and thus receives several esti-
mates of the overall residual time shift. The large number of
shifts that are then mapped to each surface position stabilizes
the residual statics results by better statistics. Within the CRS
supergathers, conventional residual statics techniques, such as
trace-to-trace correlation, are applied. The pilot trace is taken
from the CRS stack with its superior signal-to-noise ratio and
overall enhanced imaging. As in conventional residual stat-
ics calculation, the CRS approach relies on an iterative im-
provement of the actual CRS stack. The CRS residual statics
procedure was described in detail by Ewig (2003) and Koglin
(2005).

C 2006 European Association of Geoscientists & Engineers, Geophysical Prospecting, 54, 709719
CRS residual statics were calculated for the 2D seismic data
from Saudi Arabia. The CRS moveout correction that was
applied in these calculations corresponded to the stacking ve-
locity used in Fig. 7. The resulting residual statics were used to
produce the updated CMP stack of Fig. 9. Comparison with
Fig. 7 shows a smoothing of the reflection events throughout
the seismic section. The signal-to-noise ratio and the event
continuity are increased. Since the main static problems on
both sides of the section could be solved by previous con-
ventional statics approaches, the CRS residual statics mainly
affect the centre, where the steep slope down to the wadi pre-
vented shooting. The discontinuities that had remained here
largely disappeared. This is shown in more detail in Fig. 10.
The previous improvements in the CMP stack were obtained
by using parameter estimates of the CRS technique only. How-
ever, the actual CRS stacking in apertures across several CMP
locations also contributes to a significant increase in resolu-
tion and signal-to-noise ratio, as was shown for low-fold 3D
 CRS strategies for solving severe static and imaging issues 719
seismic data (Trappe et al. 2005). In the 2D data case pre-
sented here, CRS stacking incorporated the CRS-based results
obtained previously, e.g. the stacking velocities used in the
CMP stack of Fig. 7 and the residual statics applied in Fig. 9.
The resulting final CRS stack is shown in Fig. 11. It is obvi-
ous that the static and dynamic corrections are similar in the
CMP stack of Fig. 9 and in the CRS stack of Fig. 11. The
main difference between these stacks is the increased signal-
to-noise ratio in the CRS stack due to the increased CRS fold.
The whole effect of CRS processing, however, is demonstrated
by comparing the CRS stack of Fig. 11 with the conventional
processing result of Fig. 6. The CRS processing achieved a lat-
eral consistency of the signal character and a strong increase
in event continuity and resolution.
CONCLUSIONS
The seismic line considered in this case study is a typical ex-
ample of land data from an arid region. Strongly varying static
shifts from deep weathering zones pose the main problem and
the imaging quality depends primarily on the derivation of
appropriate static corrections. A straightforward but careful
derivation of refraction statics from picked first arrivals is fun-
damental to the solution of the static problem. Subsequent
iteration of manual velocity analysis and conventional resid-
ual statics, however, does not completely solve the remaining
imaging problems. In areas with strong near-surface inhomo-
geneities, e.g. below deep bottom fills of a wadi, the signal
character still differs markedly from undisturbed areas with
outcrops of consolidated rocks at the surface. The seismic sig-
nal frequency and continuity of the events is significantly de-
creased in the disturbed areas.
However, these remaining problems due to near-surface
inhomogeneities can be solved by CRS processing. The lo-
cal adaptation of CRS imaging parameters at every point of
the image automatically derives very detailed dynamic cor-
rections. These corrections produce a major increase in sig-
nal frequency and event continuity, and widely adapt the sig-
nal quality in disturbed areas towards the undisturbed signal
character. It emerges that the strong surface inhomogeneities
induce small-scale variations in the stacking velocities in the
subsurface below; these cannot be resolved by manual velocity
analysis and by the resulting smooth velocity field.
The subsequent CRS processing benefits from the large CRS
stacking fold which exceeds conventional CMP fold by ap-
proximately one order of magnitude. During the calculation of
CRS residual statics, the number of estimates for static shifts

C 2006 European Association of Geoscientists & Engineers, Geophysical Prospecting, 54, 709719
is increased correspondingly by an order of magnitude, en-
suring a much more stable averaging and convergence. The
resulting CRS residual statics generally increase the amplitude
and smoothness of the reflections and remove the remain-
ing surface-induced discontinuities. During stacking, the high
CRS stacking fold also directly improves the signal-to-noise
ratio in the final CRS stack.
ACKNOWLEDGEMENTS
We thank Saudi Aramco for permission to publish results from
their data set. We enjoyed the cooperation with Ingo Koglin
and Zeno Heilmann of the WIT consortium research group at
the University of Karlsruhe, Germany.
REFERENCES
de Bazelaire E. 1988. Normal moveout revisited inhomo-
geneous media and curved interfaces. Geophysics 53, 143
157.
Ewig E. 2003. Theory and Application of Residual Static Correction
by Means of CRS Attributes. Masters Thesis, University of Karl-
sruhe, Germany.
Gelchinsky B. 1988. The common reflecting element (CRE) method
(non-uniform asymmetric multifold system). 6th Geophysical Con-
ference, ASEG, Expanded Abstracts, 7175.
Gierse G., Pruessmann J., Laggiard E., Boennemann C. and Meyer H.
2003. Improved imaging of 3D marine seismic data from offshore
Costa Rica by CRS processing. First Break 21, 4549.
Hubral P., ed. 1999a. Macro model independent seismic reflection
imaging. Journal of Applied Geophysics 42 (3,4). Special issue on
Karlsruhe Workshop.
Hubral P. 1999b. Seismic illumination. The Leading Edge 18, 1269
1271.
Jager
R., Mann J., Hocht G. and Hubral P. 2001. Common-Reflection-
Surface stack: Image and attributes. Geophysics 66, 97109.
Koglin I. 2005. Estimation of Residual Static Time Shifts by Means of
the CRS Based Residual Static Correction Approach. Logos Verlag,
Berlin.
Mann J. 2002. Extensions and Applications of the Common-
Reflection-Surface Stack Method. Logos Verlag, Berlin.
Ronen J. and Claerbout J.F. 1985. Surface-consistent residual statics
estimation by stack-power maximization. Geophysics 50, 2759
2767.
Trappe H., Coman R., Gierse G., Robinson S., Owens M. and Mller
Nielsen E. 2005. Combining CRS technology and sparse 3D seis-
mic surveys a new approach to acquisition design, 67th EAGE
Conference, Madrid, Spain, Extended Abstracts, P182.
Trappe H., Gierse G. and Pruessmann J. 2001. The Common Reflec-
tion Surface (CRS) stack structural resolution in Time Domain
beyond the conventional NMO/DMO stack. First Break 19, 625
633.