American Oil and Gas Reporter - November 2016 - 48

SpecialReport: Oil & Gas Computing
FIGURE 2
Initial Velocity (Left) and FWI-Updated Velocities
At Different Frequencies
1,520
1,700

(m)
1,600

1,900

1,800

2,100

2,000

2,300

a)

b)

c)

d)

e)

2,500

FIGURE 3
Modeling with Initial Velocity Model (Left)
And FWI-Updated Velocity Model (Right)
(ms)
4,000

(ms)
4,000

5,000

5,000

a)

Offset: 4,500 m - 7,200 m

b)

Offset: 4,500 m - 7,200 m

FIGURE 4
Gathers from Reservoir Location Before (Left) and After (Right) FWI

Offset
320 m

6,620 m

level and more closely matched the check
shot data.
Encouraged by the comparison, multiple FWI iterations were performed from
3.0 to 18.0 hertz using both diving wave
and reflection data. The inversion began
with low frequencies to reduce the likelihood of cycle skipping, and the frequency
was increased gradually in an increment
of 1.0 hertz for each step.
More details were added progressively
to the velocity model as the frequency increased, as shown in Figure 2, which begins
with the initial velocity (panel A) and progresses through FWI-updated velocities at
48 THE AMERICAN OIL & GAS REPORTER

different frequencies (panels B-E).
As shown by the red arrows, slow
velocities corresponded to the strong
amplitude events within the MTC, with
higher velocity observed in a shallow
channel. Identifying such shallow velocity
anomalies can help pinpoint drilling hazards.
Analysis Of Results
To validate the FWI updates, the results
were examined in both the data and image
domains. Data-domain analysis compares
the modeled shot gathers with the recorded
data. When overlaid, coinciding wavelet

peaks indicate a close match between
the modeled and recorded data.
The modeled data were displayed in
color with positive amplitude as red and
negative as blue fill, while the recorded
data were in wiggle with positive amplitude filled in with black. The modeled
shot gather using the FWI-derived velocity
matched the recorded data better than
the initial, tomography-derived model.
Figure 3 displays the recorded shot
gather (with an 18-hertz high-cut filter) in
wiggle with positive amplitude filled in
with black, overlaid by modeled shot
gathers in color with positive amplitude as
red and negative as blue fill. Modeling
with the initial velocity model is shown on
the left in Panel A, while modeling with
FWI-updated velocity model is at right.
The modeled data matched the recorded
data better with the FWI model, with
peaks of the recorded data (black wiggles)
covering peaks of modeled data after
FWI, as evidenced by less visible red
after FWI than before.
A Kirchhoff migration also was performed to evaluate the results in the image
domain. Deghosting was applied to improve
the resolution of the image before migration,
along with "designaturing" and "debubbling." In the migrated stack image from
the FWI-derived velocity model, the events
below the MTC were less distorted and
more focused, and the structures became
simpler and more continuous.
The reservoir event at the bottom of the
section (approximately 3,200 meters) also
became more continuous after FWI. Gather
flatness below the shallow velocity anomalies improved, especially the nonhyperbolic
residual move-out at reservoir depth, which
is difficult for curvature picking and inversion in ray-based tomography.
With the improvement in the stacked
image and in the gather flatness, more
realistic amplitude behavior was expected,
including amplitude versus offset. Amplitude was measured at the reservoir on
migrated gathers (within the red dashed
lines in Figure 4), with amplitude plotted
in the blue lines above both before (left)
and after (right) FWI.
Before FWI, there is an abrupt amplitude increase at offset 1,820 meters and
a decrease at offset 3,620 meters. On the
other hand, the AVO trend became
smoother after FWI, which can lead to
more stable AVO modeling analysis.
The analysis also interpreted the top
of the reservoir horizon on the migrated
stacks (with and without FWI), and ex-
American Oil and Gas Reporter - November 2016

Table of Contents for the Digital Edition of American Oil and Gas Reporter - November 2016
