Summary
GrAZ 3D is a volume based method for fluid contact
detection and anomaly assessment. It is
the extension to 3D volume analysis of the corresponding horizon based 2D GrAZ methodology.
In a 3D GrAZ calculation, a dot product is formed between a
structural dip vector field and a gradient vector field of an attribute
volume. The result is a 3D volume of dot
products after final filtering. While
the horizon based method is better constrained, the requirement that a horizon
be interpreted in advance limits its applicability to anomaly assessment. The 3D GrAZ methodology has broader applicability, but as an
effective derivative of an attribute volume, the artifact load is high and
extensive post filtering is necessary.
Given the existence of paleo fluid contacts, in addition to data
artifacts, a GrAZ 3D anomaly is best viewed as a necessary but not a
sufficient indicator in an exploration workflow. The calculation can be done in a scanning
mode useful for anomaly identification before detailed horizon
interpretation.
Introduction
Before beginning the
process of detailed structural or amplitude interpretation, it is useful to
have indicators that lead an interpreter to focus in certain areas. GrAZ 3D is designed to glean firstpass indicators of
hydrocarbon fluid contacts, and thus provide leads to regions deserving more
intensive work. GrAZ 3D is a straightforward extension of the horizon
based equivalent but without the need of a previously interpreted horizon.
The problems of
converting a 2D horizon process to a 3D process that can be run in a scanning
mode are many. The existence of an
interpreted horizon greatly increases the legitimacy of subsequent
calculations, having already had both geology and interpretive noise rejection
incorporated into the horizon itself. In
proceeding to a 3D scanning implementation, a goal is to incorporate some
geologic reasoning and interpretive noise rejection in the computation
processes.
2D GrAZ calculation
2D GrAZ is a method
of determining whether changes in a derived horizon attribute change abruptly
in the updip direction at a constant time or depth and, hopefully, due to a
fluid contact. A typical horizon based attribute
might be the maximum instantaneous amplitude near the picked event or simple
maximum negative trough amplitude. To
make this attribute change determination, the gradients of both the interpreted
Z values and of the extracted attributes are computed in 2D. This produces 2 gradient vectors at each
horizon grid location. A dot product is
formed using both vectors. The dot product can be put through a Horizon Binning
process to make a graph like the one shown in Figure 1, or can be mapped
directly, as in the example of Figure 2.
The goal of GrAZ 3D is to duplicate a map, like Figure 2, but without
needing to interpret a horizon in advance.
Figure 1: In an 2D GrAZ plot after Horizon Binning, the Xaxis is time or
depth and increases to the right. The vertical axis is the binned dot
product of the gradient of the
structure map with the gradient of the seismic attribute. In the waterleg,
isotropy is inferred by a very small dotproduct values to the right, down
structure..

Figure 2: A 2D GrAZ dot product can be mapped. Shown is a 2D GrAZ calculation with contours overlain. The location
of the anomaly (roughly northsourth) corresponds with an inferred fluid
contact. Downdip is to the northeast. Note that the anomaly
appears to break at likely fault locations.

GrAZ 3D
calculation: 3 steps
Following the basic
concept of a 2D GrAZ calculation, there are 3 steps in computing a GrAZ 3D volume.
First, a volume of structural dip vectors must be computed. Second, a dot product of the gradient of the
attribute volume with corresponding structural dip vectors must be
performed. Third, since the artifact
load of this sort of derivative process is high, a sequence of filters is
normally applied before loading onto a workstation (or Segy clustering to horizons). These steps are described in more detail in
reference 3.
Structural dip
vector volume
There are many good
methods available for computing dip vectors using 3D seismic data. Several are described by Marfurt in reference
2. In a GrAZ 3D calculation, the dip vectors must function as the
gradients of “horizons,” in analogy to the 2D GrAZ calculation.
This means that the dip vectors must indicate the overall dip in an
“upstructure” direction, in a manner that is stable and geologically
reasonable. This author has found that
the dip vectors taken from a standard calculation are too unstable to use for
fluid contact detection without further processing. In particular, variations in phase near fluid
contacts can cause dip vector variations in the exact location where one needs
stability prior to taking the dot products.
For this reason, from a normal dip vector calculation, “structural dip
vectors” are extracted which indicate the overall structural dip, focused on
locations that are likely to have stable values. These structural dip vectors are used in
computing the GrAZ 3D dot product.
Dot product and
artifacts
It is
straightforward to compute a gradient of a 3D attribute volume and then
compute a dot product with appropriately selected structural dip vectors. This is effectively a directional derivative
of the seismic attribute in an updip direction. In the water leg of a reservoir unit, one
assumes that changes will be isotropic and cause small, randomlyplaced dot
product values. This allows the
inference that large dot products at a consistent time (depth) are related to
fluid contacts, if other causes can be discounted by filtering or later
interpretation.
Final filtering
and interpretation
As effectively a
directional derivative of seismic data, there are many things that can give
large values. Faults, salt faces and
rapid stratigraphic changes can all produce large derivative values. In addition, paleo fluid contacts exist and
can cause significant derivative values.
Paleo fluid contacts mean that “legitimate fluid contacts” may be
detected at locations where uneconomic wells would be drilled. This means that the existence of a GrAZ 3D anomaly is not sufficient to assert hydrocarbons
in an exploration prospect analysis.
Conversely, the lack of a GrAZ 3D anomaly, at a prospect whose petrophysics would
infer one, is very negative for the prospect.
Normally, before
loading to a workstation, a sequence of final filters is applied. These are designed to reduce randomly large
derivatives, possible faults, and other features that one would not normally
consider to be associated with a legitimate fluid contact. Figure 3 shows a time slice through a GrAZ 3D volume after a full sequence of filters. The “white areas” in this timeslice are
“0.0.” Interpretation is normally done on time slices, since on a vertical
section the anomalies may appear as small “dots.” In a time slice, the “dot’ may have extent
and extend across a fault block. On a
workstation like SMT or Geoframe, the GrAZ 3D anomalies can be picked as faults (in time slice)
and then incorporated into the rest of the interpretation. In any project involving GrAZ 3D work, one finds that the calculations are
relatively fast and serve as a starting point for more extensive work to
follow.
Figure 3: A GrAZ 3D anomaly associated with the
fluid contact for the perf shown. The anomaly basically spans one fault
block. It can be picked as a fault on
a workstation or extractged as a horizon using a clustering algorithm.that
the anomaly appears to break at likely fault locations.

Conclusions
The problems of
converting a 2D horizon process to a 3D process that can be run in a scanning
mode are many. The existence of an
interpreted horizon greatly increases the legitimacy of any subsequent
calculation, having already incorporated both geology and interpretive noise
rejection. In proceeding to a 3D
scanning implementation, geology and interpretive noise rejection will come
later, but the need to preinterpret horizons is removed.
GrAZ 3D detects
locations of fluid contacts when there is a change in a seismic attribute from
downdip to updip, emphasizing constancy in time or depth. As a derivative based operation, there are
many artifacts that must be removed before or during the interpretation
process. Some of the artifacts are due
to paleo fluid contacts, these cause GrAZ 3D anomalies to be viewed as necessary but not
sufficient, in an exploration setting.
In any project
involving GrAZ 3D work, the calculations are relatively fast and
serve as a starting point for more extensive work to follow. This technique is not a “silver bullet,” but
can logically augment a normal exploration, delineation and/or development
workflow in many settings around the world
Acknowledgements
The author would
like to thank OpenSpirit Corporation for permission to use the data analyzed in
Figure 2 and to thank Centerline Geophysics for permission to show the results
in Figure 3.