Anisotropic prestack depth migration improves the well-ties at ... - PGS

Anisotropic prestack depth migration improves the well-ties at the Ewing Bank Oil Field, Gulf
of Mexico
Dario Cegani, ENI E&P, Eduardo Berendson, ENI Petroleum, Clive Hurst, ENI Petroleum, Helen Delome*,
PGS Geophysical, USA and Rubén D. Martínez, PGS Geophysical, USA
Summary
We successfully applied anisotropic Kirchhoff prestack
depth migration (AKPreDM) to a 3D data set from the
Ewing Bank Oil Field, Gulf of Mexico to accurately map
the target sands in depth. Initially, we used isotropic
Kirchhoff prestack depth migration (IKPreDM) to perform
the mapping of the reservoir sands. However, the depths
estimated using the seismic data and those detected at the
wells for the target sands showed excessive miss-ties
(approximately 1100 feet). The results obtained after the
application of AKPreDM showed that the miss-ties were
greatly reduced within a few tens of feet, therefore
providing a better tie with the well tops. In this paper, we
present the methodology employed in this case study,
which includes the anisotropic parameter estimation. We
also discuss the well tie results obtained after the
application of AKPreDM and IKPreDM. We conclude that
the effect of anisotropy (vertical transverse isotropy, VTI)
should be taken into account to obtain seismic depth
estimates that better tie the well depths accurately.
The field trap is formed by the up -dip stratigraphic
pinchout of turbidity sands onto a salt-supported structural
ridge. Sand flows were funneled generally from the westnorthwest
and encountered a salt barrier trending
approximately north south. Environments of deposition of
these sands are amalgamated to channelized basin floor
fans and channel complexes all confined in an intra-slope
mini-basin. Pay sands are at or below tuning thickness,
with occasional amplitude masking by overlying events
(Fig. 1).
Introduction
Lately, the existence of anisotropy (vertical transverse
isotropy, (VTI)) in velocities measured from seismic data
has been widely recognized. The presence of anisotropy
can cause depth errors if the depth imaging process is
performed using isotropic assumptions. Our experience in
the Ewing Bank Oil Field from the Gulf of Mexico
revealed that isotropic depth imaging yielded significant
depth miss-ties with the wells. The VTI effect is believed to
be related to the overburden thin layering. Thin layering
whose layer thicknesses are smaller than a seismic
wavelength will induce VTI. By performing anisotropic
depth imaging, we observed a significant quantitative
reduction in the miss-ties between the seismic depths and
the well depths. An anisotropic P-wave imaging strategy
similar to that presented by Martinez and Lee (2002) was
used for this purpose. In this paper, we present the imaging
methodology used in the Ewing Bank Oil Field. The
geologic background of this Field is given in the next
section.
Field Overview
The Ewing Bank Oil Field is located in 1,700 feet of water
and development drilling started in 1998.
Fig.1: Schematic representation of reservoir depositional Model
Salt-related faults in the field area are also important to
predict reservoir continuity and to explain the interaction
between salt movement and sedimentation. Geologic field
data consists of 3D seismic and 13 penetrations on which
hydrocarbon-bearing sands are clearly detected (Fig.2).
Reservoirs in the field area are between 13,000 and 15,400
feet. Sands are of excellent reservoir quality at all levels,
with porosities ranging from 27 to 31 percent, net -to-gross
ratios typically of 80 percent or more, and water saturations
from 10 to 40 percent. Permeabilities of 100’s of milldarcy
for the shallow reservoir sands are common, while deeper
sands are generally over one Darcy. Oil gravity in the field
varies from 25 API in the shallower sands to 30 API in the
deeper sands. The oils contain 1.6 percent sulfur.

Anisotropic prestack depth migration improves the well-ties at the Ewing Bank Oil Field, Gulf of Mexico
simultaneously with the η eff (Min et. al., 2002). We
proved that smoothing the η eff model provided more
stability. After some testing, the final η eff model was
determined by averaging the interval η values around the
#1 well. We used only one interval η defined in one layer.
The layer was defined between the water bottom and a
surface above the reservoir, and the interval η value was set
to 0.012. This surface was interpreted (from the isotropic
final stack volume) as the bottom of the sand/shale
sequence in the area (see Figs.3 and 4). We concluded that
this layer was the main cause of the VTI effect.
Fig.2 Salt mini-basins in field area. In blue (dark) is the
spatial distribution of a key reservoir in the field. Yellow
lines show some of the wells in the field.
η=0.012
Depth imaging in the presence of anisotropy
We first performed isotropic depth imaging resulting in
common image gathers (CIGs) that were nearly flat in
depth, but the seismic depths obtained at the reservoir
levels were off to those detected at the wells. The miss-tie
estimated at the oil field was about 1100 ft. This prompted
us to investigate the possibility of having velocity
anisotropy effects in the area.
To test the hypothesis of having velocity anisotropy, we
used the anisotropic P-wave imaging strategy proposed by
Martinez and Lee (2002). The most difficult part in
anisotropic imaging is the estimation of the parameters that
describe the vertical transverse isotropy (VTI). Three
parameters are necessary to define a VTI medium: interval
seismic velocity; interval vertical velocity; and the interval
anelliptical parameter η.
Fig.3 The one layer η eff model.
Interval seismic velocity model building: The isotropic
final sediment velocity model was used as the initial
seismic velocity model for the anisotropic depth imaging
process.
h model building: The CIGs were stretched to time and the
isotropic interval velocity model was converted to RMS
velocity in time to pick the effective η. The effective η
functions were converted to interval η in depth using the
Dix type formula given in Alkhalifah, T., 1997, which uses
the corresponding velocity model.
Many tests were conducted to create a stable η eff model.
The effective RMS velocities were determined
Fig.4 Depth horizon map for the base of the
sand/shale sequence.

Anisotropic prestack depth migration improves the well-ties at the Ewing Bank Oil Field, Gulf of Mexico
Interval vertical velocity model building: The vertical
velocity model was computed from the interval seismic
velocity model and the interval vertical velocities measured
in wells using VSP check shots. We transformed both
velocities from interval to RMS velocities. We then used a
linear relationship assumption between the two velocities,
i.e., V RMS-vert = A + B*V RMS-seis , where the coefficients A
and B are constant. The A and B coefficients were obtained
from the cross-plots of the RMS seismic velocities in time
at well locations and the RMS vertical (well) velocities in
time computed from the VSP check shots. Two of the
straightest wells: #1 and #2 were used for this analysis.
Both wells were used to compute A and B at the two well
locations. For the #1 well, A=395 and B=0.75 (see Figure
5-a). For the #2 well, A = 425 and B=0.727 (see Figure 5-
b). An anchor was put at the location (2138, 6660) for zero
anisotropy effect (A=0, B=1). Volumes of A and B (A(x,y)
and B(x,y)) were computed by triangular interpolation of
the 3 values at the 3 locations. Thus, V RMS-vert in time was
computed and converted to interval velocity
Vwell-rms (m/s)
2500
2000
1500
1000
500
#1 Well
0
-500 0 500 1000 1500 2000 2500
V1-rms (m/s)
in depth and minor smoothing was applied. The water layer
was then re-inserted in the vertical interval velocity model.
Having the interval seismic velocity, interval η and the
interval vertical velocity models, the next step in
AKPreSDM was to perform anisotropic ray tracing for the
anisotropic traveltimes calculation. The traveltimes were
then used in anisotropic Kirchhoff prestack depth migration
(AKPreSDM). New CIGs were obtained to perform the
update of the anisotropic parameters for the next iteration.
The depth sections and gathers were used for QC and welltie
analysis. The depth miss-tie was small at the central and
north of the field (+41 to +65 ft at the #1 well, and -49 to –
39ft at the #2 well). Meanwhile in the Southern part of the
field, the miss-ties were larger (about -175ft).
Vrms
Linear (Vrms)
Fig. 5-a Cross-plot of RMS vertical (well) velocities
versus RMS seismic velocities.
Vwell-rms (m/s)
2500
2000
1500
1000
500
# 2 Well
0
-500 0 500 1000 1500 2000 2500
V1-rms (m/s)
Fig. 5-b Cross-plot of RMS vertical (well) velocities
versus RMS seismic velocities.
Based on these results, another velocity refinement step
was undertaken with the objective to reduce the well depth
to seismic depth miss-tie.
The updated velocity functions were re-gridded and
properly smoothed to generate a new interval seismic
velocity volume. Cross-plots of the new RMS seismic
velocities versus the RMS vertical velocities at the two well
locations were re-computed to generate the new linear
relationship. For the #1 well, the relation was very close to
the previous iteration (A=400 and B=0.75), but for the #2
well a slight change was needed (A = 415 and B=0.739).
The anchor at (2138, 6660) for zero anisotropy was still in
effect (A=0, B=1). New volumes of A and B (A(x,y) and
B(x,y)) were computed by triangular interpolation of the
three values at the three locations. A new interval vertical
velocity in depth (V vert ) was re-created in the same way as
the previous iteration. The interval η model was kept
unchanged.
Anisotropic Model Verification Tests
To prove the accuracy of the models, a model verification
test was conducted to output PSDM gathers for 10 inlines
and 3 cross-lines: inlines 1926, 1936, 1938, 1942, 1944,
1946, 1950, 1954, 1964, and 1978, cross-lines 5892, 5896,
and 5900. The gathers and stacks were used for a well-tie
analysis. We found that the miss-tie away from the #1 well
(especially in the Southern part of the field) was improved
while at this well location the miss-tie was slightly
increased (~+100ft). The gathers were acceptably flat. We
performed additional analyses to understand the possible
factors that would have affected the depth ties and to
propose solutions for better depth ties without degrading
the image quality. We found that the VSP check shot of the
#2 well was not useful because there were measurements
only between 3.9 seconds to 4.5 seconds (below the
sand/shale sequence). So the vertical velocity model was
Vrms
Linear (Vrms)

Anisotropic prestack depth migration improves the well-ties at the Ewing Bank Oil Field, Gulf of Mexico
re-built by using only the linear relation from the #1 well.
This relation was re-computed by using the measurements
above the reservoir (3.8 seconds and at earlier times). The
new values of A= 385 and B=0.75 were obtained from a
new cross-plot.
The PSDM gathers for the same 10 inlines were output for
this new vertical velocity model. The stacked sections
showed the best depth ties at the well locations. The gathers
were reasonably flat. Figures 6 and 7 show portions of the
migrated stacked sections at the locations of wells #1 and
#2 respectively.
Fig. 7 Comparison of isotropic (right) and anisotropic
(left) depth imaging results at the #2 well location. The
isotropic PSDM image (right) was shifted 1100 ft to try
to tie the well.
Fig. 6 Comparison of isotropic (right) and anisotropic
(left) depth imaging results at the #1 well location. The
isotropic PSDM image (right) was shifted 1100 ft to try
to tie the well.
Conclusions
Through this case study, we demonstrated that velocity
anisotropy significantly impacts the seismic depths
estimated using Kirchhofff prestack depth migration in the
Ewing Oil Field.
When the depth migration process is performed with
corrections for the VTI effect, the miss-ties with the well
tops are greatly reduced. The image quality was, in general,
improved, e.g. see figure 7.
As a result of the anisotropy corrections, the structural
behavior of the reservoir sands changed with respect to that
obtained with isotropic depth migration. To validate the
structural results from both isotropic and anisotropic depth
migrations, dipmeter logs may be used to compare the dips
estimated from the seismic data with those of the dipmeter
log.
References
Alkhalifah, T., 1997, Velocity analysis using nonhyperbolic
moveout in transversely isotropic media, Geophysics 62,
1839-1845.
Martinez, R.D. and Lee, S., 2002, A strategy for anisotropic
P-wave prestack imaging, 72 nd Ann. Internat. Mtg., Soc.
Expl. Geophys., Expanded Abstracts, 149-152.
Acknowledgments
The authors of this paper would like to thank ENI
Petroleum for granting permission to publish this paper.
We also thank ENI Petroleum for providing the oil field
background information and migrated data sets to illustrate
this paper.