American Oil and Gas Reporter - September 2015 - 89

SpecialReport: Reservoir Characterization

Broadband 3-D Characterizes Reservoir
By Peter R. Mesdag
and Menne D. Schakel

HOUSTON-Broadband acquisition
and processing technology has revolutionized seismic acquisition and interpretation. The revolution is not yet complete, however, as earth scientists and
petroleum engineers continue to explore
how broadband can affect reservoir characterization efforts. In particular, the ultimate benefit of broadband seismic will
be meaningful reservoir properties with
unprecedented bandwidth and accuracy
through simultaneous amplitude-variation-with-offset inversion.
Broadband technology has a wider
seismic spectrum with energy at lower
frequencies than conventional seismic.
Enhanced high-frequency content leads
to a sharper wavelet and higher-resolution
images. Enhanced low-frequency content
reduces wavelet side lobes, may eliminate
the need for a low-frequency model from
interpolated well logs, and results in
better signal penetration.
The impact of broadband seismic on
structural interpretation already has been
demonstrated to be dramatic. No less of
an impact should be expected from broadband inversions for reservoir characterization. The impact of the low frequencies,
in particular, is illustrated in Figure 1,
where a stacked sandstone (impedance)
model has been filtered to show the effects
of the low frequencies below the band of
traditional acquisition and processing. By
adding low frequencies, the impedance
values converge toward the true profile.
When performing inversion on broadband seismic data there are particular
properties that must be considered in different steps of the process. In addition,
careful thought has to be given to how to
use inversion quality controls (QCs) to
indicate the value of broadband seismic
in the inversion stage.
Wavelet estimation is critically important to inversion processes. Broadband,
with its extended low-frequency range,
demands longer wavelets, and therefore,
longer wavelet design windows. Innovative
techniques are necessary to estimate the
low-frequency components of these
wavelets while retaining relevance to the
reservoir in the high-frequency band.
These issues were addressed in a
project that performed a full-stack im-

pedance inversion of broadband data from
a multiclient seismic survey on the U.K.
Continental Shelf. Seismic data were inverted using a low-frequency model (0-2
hertz) derived from seismic velocities
and a broadband wavelet constructed by
merging both a high-frequency seismicdominated wavelet and a low-frequency
wavelet from a longer time window.
Sample results are shown in Figure 2.
Note that the limits of the channel are
much better defined and "geologic-looking" in the broadband inversion (right)
compared with the conventional inversion
(left). Lithofacies derived from inversions
also showed more spatial heterogeneity
since they were not constrained by lowfrequency models from logs.
Seismic Inversion
For accurate reservoir characterization,
the seismic data need to be converted to
meaningful reservoir properties. An intermediate step toward reservoir properties
is through elastic parameters, which are
layer properties that are typically combinations of compressional (P) velocity,
shear (S) velocity, and density. Seismic
inversion is the process of converting
seismic data to elastic parameter data.
There are several key components to

seismic inversion, including:
* The wavelets;
* The low-frequency trend or starting
model; and
* The inversion engine.
Every seismic volume has its own
wavelet. Processing takes great care to
produce a zero-phase wavelet. However,
residual corrections are commonly applied
to it when inverting the seismic. Wavelets
for seismic inversion need to be accurate
both in amplitude and phase for all frequencies that are in the seismic data. Estimating wavelets for broadband inversion
is more challenging than for conventional
seismic inversion, but they can be estimated in current software packages.
In regard to the low-frequency starting
model, the frequency band needs to be
extended to zero Hz on the lower end of
the seismic spectrum in order to convert
seismic data to layer properties. By nature
of the physics, this cannot come from
the seismic data alone. So any inversion
needs low-frequency information to come
from other sources. The requirements for
the low-frequency trend are much less
demanding for broadband data. Consequently, the seismic inversion process is
driven more by the data.
Inversion algorithms use assumptions

FIGURE 1
Stacked Sandstone Sequence Filtered to Show
Lowest Recoverable Frequencies from Broadband
Model

Model (notch 1-6 Hz)

Frequency
(Hz)

Frequency
(Hz)
0

Model (notch 1-2 Hz)

0 1

Frequency
(Hz)
01 2

SEPTEMBER 2015 89

American Oil and Gas Reporter - September 2015

Table of Contents for the Digital Edition of American Oil and Gas Reporter - September 2015