The seismic interpreter now has the
use of a simple tool to aid in the search for hydrocarbons by allowing
individual target reflection events to be spectrally analyzed and
compared to gas response modeled from well logs with and without
pay. The seismic reflection is then decomposed to spot a positive
gas response.
Until now, spectral decomposition techniques have used windowing
methods to decompose the seismic trace into its constituent frequencies
— but these methods:
- Mix reflection events and introduce unwanted
artifacts into the data.
- Restrict the usefulness of spectral decomposition
to the inspection of single-frequency maps to try to relate amplitude
maxima to geologic events.
Recent advances in techniques have yielded a method that does
not use windowing to decompose the trace: The Instantaneous Spectral
Analysis (ISA) method uses a wavelet transform technique to produce
single-frequency reflection events that are accurately localized
in time.
Each full-spectrum reflection can be visualized and analyzed at
its uncontaminated single-frequency equivalents.
Frio Seismic Bright Spot
Figure 1 shows a high amplitude reflection
characterizing a Frio reservoir trapped stratigraphically as a sand
pinchout.
The Frio sand is about 68 feet thick, and is shown in figure
2 with the well log synthetic seismogram tie. Notice that the
gas pay has a low velocity compared to the brine-filled part of
the sand at the base. This adds significant strength to the reflectivity
of the sand body, causing it to be seen as a high amplitude reflection;
the classic bright spot.
What cannot be seen is the behavior of the individual seismic
frequencies; i.e. what effect does the hydrocarbon charge make on
the amplitudes of each discrete frequency. Because the ISA technique
allows uncombined reflectivity to be examined, as no windowing is
used during the calculation, the pay reflectivity can be isolated
and studied.
This new approach allows one to show the reflection's response
to the hydrocarbon charge at various frequencies via a "frequency
gather," as shown in figure 3a. The
display shows increasing frequency to the right with the strongest
amplitudes in warm colors.
This is very similar to the familiar AVO gather — except where
adjacent traces represent the reflection's response to changing
offset in the AVO gather. Here each trace represents the reflection's
amplitude at a single frequency, or amplitude versus frequency (AVF).
The anomalous response caused by the pay clearly can be seen as
a very high amplitude with a peak frequency that is shifted toward
the high end of the useable bandwidth. When the process is run on
the entire seismic line, single-frequency panels are produced as
shown in figures 3b and 3c.
Note that at 10 Hz, the pay does not exhibit high amplitude, while
at 36 Hz, it is one of the brightest events on the section.
The Frio bright spot on the 36 Hz seismic line in figure 3c agrees
with the frequency gather shown in figure
3a. The pay has relatively little energy at 10 Hz, but at 36
Hz, it is one of the few remaining events to have high reflection
strength. This is in contrast to the strong events centered between
2.0 and 2.1 seconds at the wellbore. They have visually lower frequency
and their strongest reflection amplitudes are closer to 10 Hz.
When viewed as a frequency panel movie, the changing contrast
becomes very striking.
3-Dimensional Map Displays
Now let's look at the data in 3-D.
When the ISA process is applied to cubes of seismic data, the
results are a series of single-frequency cubes that are loaded onto
the workstation and interpreted.
Figure 4 shows four slices from the
frequency cubes on the pay horizon:
- The first is at 24 Hz, that frequency which
has a minimum amplitude response for the area surrounding the
pay on strike (indicated by the arrows).
- The second is at 32 Hz ,or near the maximum
of the amplitude response at the pay.
- The third is at 47 Hz, which shows a minima
in the amplitude spectra of the pay as seen on the frequency gather
at the arrow.
- The fourth at 58 Hz shows the reflectivity
of the pay close to background.
As figure 4 shows, the pay is acting
completely different than the surrounding sand when viewed at discrete
frequencies. This is even more apparent when all the frequency maps
are viewed as a movie. The pay has a distinctly different dynamic
frequency response than the background because the hydrocarbons
have changed the reflectivity of the reservoir.
Well Log Modeling Confirms Frequency Response
To understand the seismic response, let's examine the detailed
reflectivity obtained from well logs.
Figure 5 shows the modeled response
using sonic and reflectivity logs, which explains this difference
in dynamic behavior. The only change between the two curves is that
the velocity of the Frio pay zone has been replaced by a brine-filled
sand velocity.
The local reflectivity of both cases has been analyzed for spectral
content and is shown in the graph of amplitude vs. frequency. One
can clearly see that the hydrocarbons are responsible for the high
amplitudes at and around 32 Hz, and the associated dimming at 47
Hz. They also are responsible for subtle changes in reflectivity
at other frequencies.
Similarly, the amplitude low at 24 Hz in the curve with no hydrocarbons
can be seen in the maps in the area surrounding the reservoir.
Spectral Results
The sand that traps this Frio field is present along strike, pinches
out updip and is not present downdip. If the observed anomalous
reflectivity were due to the sand thinning, then sequential frequency
maps should show a feature "walking" away from the field, and this
is not seen. The amplitude maxima of the reservoir at 32 Hz and
the following minima at 47 Hz, plus the amplitude minima at 24 Hz
in the brine-filled area adjacent to the reservoir observed in the
maps are explained by the spectral modeling.
There could be other geologic conditions that would cause the
reflectivity of this reservoir to more closely resemble the brine
case. A decrease in porosity, for example, will bring the reservoir
velocity closer to that of a brine-filled sand.
In the case of very little porosity, the velocity of the brine
and hydrocarbon-filled sand will be much closer together, and the
difference in reflectivity will be much smaller. Thus, the pay would
be harder to discriminate spectrally.
The technique illustrated here will work best in sands with high
porosity and permeability, but has been employed successfully in
consolidated sands and carbonates in a variety of depositional environments
and depths.
Others uses include:
- The display of attenuation and low frequency
shadows for direct hydrocarbon indication.
- The analysis of subtle thickness or porosity
changes, which result in tuning frequency changes.
Since the input to this process is simply the migrated data, the
better the data quality, the more accurate the results of this method.
Conclusion
A new type of spectral decomposition has been shown to be useful
as a simple tool to isolate the reflectivity of hydrocarbons in
a Frio sand reservoir using migrated data.
By viewing frequency maps as a movie, subtle changes in frequency
become dynamically visible. The observed unique reflectivity
of the reservoir due to the presence of hydrocarbons has been confirmed
with its theoretical reflectivity calculated from well logs.
The ISA method of spectral decomposition does not mix the reflections
in time, thus allowing the investigation of reflectivity from individual
seismic events.
This method shows great promise to become another valuable seismic
detection tool in the search for hydrocarbons.