American Oil and Gas Reporter - November 2015 - 51

SpecialReport: Oil & Gas Computing

Approach Enhances Seismic Imaging
By David Chase,
Yuval Serfaty,
Igor Belfer
and Zvi Koren

HOUSTON-Seismic imaging techniques use background velocity models
to migrate recorded seismic data from
the acquisition time domain to common
image gathers (CIGs) in the depth domain.
Normally, these are of a lower dimensionality than is potentially accessible,
with the extra dimensions being reduced
by integration. As a consequence, much
information that could be of use in uniquely characterizing subsurface geological
features is irretrievably lost. In order to
retain this important information, the imaging can be performed in two independent
stages.
The first stage involves ray-based
mapping of the seismic events into subsurface local angle domain (LAD) tables
at each image point, retaining all the relevant degrees of freedom. The LAD table
is defined by the reflection angles and
their azimuths, in addition to the two
polar angles for the ray-pair directivity.
Together with the depth axis, this results
in a 5-D LAD CIG. These are organized
and stored in an in-line/cross-line lateral
grid.
The second stage involves implementing different types of dedicated imaging
operators that work on these full-dimensional datasets to isolate desired subsurface
characteristics, such as structure continuity,
discontinuous objects aligned near vertical
planes such as salt flanks or faults, and
even fracture systems. The method is
demonstrated on a real dataset
Seismic migrations use the recorded
data and an approximated background
velocity model to back-propagate the
many wave phenomena present in the
recorded data into their "true" subsurface
locations, and by applying the "right"
imaging condition criteria, form the best
possible image of the geological model.
Among these wave phenomena, there
are those of crucial interest to the interpreter, such as reflections, different
types of diffractions, and duplex/corner
waves.
There are many challenges in this
process. The background velocity model
is a simplistic realization of the true geological model that can include, for ex-

ample, complex structural geometries and
layer velocities with different orders of
anisotropy, absorption and dispersion.
An accurate geological model builder
and appropriate wave/ray-based modeling
are essential for simulating the wave
propagation in such complex models.
The imaging condition applied at the
subsurface point is a critical operation
that strongly affects the type and class of
images obtained. Additionally, the acquisition affects the ability to optimally
illuminate the subsurface grid points from
all directions.
In seismic imaging, the recorded seismic data are generally migrated directly
to depth-domain CIGs, which can be
two-dimensional (for example, depth versus angle or offset), or more recently,
three-dimensional (with the addition of
azimuthal dependence). Because these
gathers are of a lower dimensionality
than is accessible during migration, the
migrated physical events, having been
potentially isolated in the imaging process,
are rendered inseparable or even invisible
in the final image.

Novel Approach
The novel approach introduced in
this article retains the integrity of the
collected data in a form that can be of
use in achieving an unambiguous interpretation of the subsurface geology. The
input seismic data are first mapped into
the full-dimensional decomposition for
each imaging point. This consists of
four components of the local angle domain. Dedicated imaging operators then
can be applied to form various desired
classes of final images. This imaging
approach attempts to confront the many
geological and geophysical imaging chal-

lenges encountered worldwide, and is
an extension of ongoing work in imaging
in the LAD.
The method consists of two independent stages. The first stage involves raybased data mapping of the surface seismic
data into the full subsurface space-angle
domain. A point diffractor ray tracing
operator has been designed that shoots
rays from the imaging point equally in
all directions, and stores the required ray
properties for all of those that succeed in
reaching the surface. The aim of this step
is to optimally associate the source and
receiver locations with the appropriate
subsurface directivities.
The permutations of the individual
diffracted rays form a system of reflection
ray pairs (incident and scattered) that
enable the decomposition (binning) of
the migrated seismic events into the in
situ 4-D LAD table at each subsurface
point.
The in situ 4-D LAD table comprises
two polar angles representing the directivity of the ray pairs (the sum of the incident and scattered slowness vectors)
and two additional angles representing
the opening angle and opening azimuth
between the two slowness vectors at the
image points. This type of data is stored
as a 5-D CIG for each in-line/cross-line
location. The size of the bins in the 4-D
LAD table determines the resolution of
the mapped input data. This subsurface
domain also is the most appropriate from
the physical point of view for many data
assessment and correction operations normally performed in the input data domain.
'Specularity' Gathers
The second stage involves processing

FIGURE 1
Mapping Schema from Surface to Subsurface Domain (A), 5-D Gather
(B), Opening Angle (C) and Directional Angle (D) 3-D Gathers

NOVEMBER 2015 51

American Oil and Gas Reporter - November 2015

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