Unconventional/complex reservoirs

Unconventional-Reservoir Characterization With Azimuthal Seismic Diffraction Imaging

The complete paper proposes an azimuthal plane-wave-destruction (AzPWD) seismic-diffraction-imaging work flow to efficiently emphasize small-scale features associated with subsurface discontinuities such as faults, channel edges, and fracture swarms.

Fig. 1—Edge-diffraction schematic. Bottom: seismic acquisition over the channel; dashed lines: edge-diffraction ray paths; triangles: sources and receivers combined; top: corresponding edge-diffraction travel-time surface (“Parallel,” “Perpendicular,” and “Intermediate” denote PWD directions in relation to edge-diffraction orientation).

The complete paper proposes an azimuthal plane-wave-destruction (AzPWD) seismic-diffraction-imaging work flow to efficiently emphasize small-scale features associated with subsurface discontinuities such as faults, channel edges, and fracture swarms and to determine their orientation by properly accounting for edge-diffraction phenomena. The work flow is applied to characterize an unconventional tight-gas-sand reservoir in the Cooper Basin in Western Australia. Extracted orientations of edges provide valuable additional information, which can be used by the interpreter to locate finer-scale features and distinguish them from noise.


Unconventional reservoirs may exhibit high structural variability, which is difficult to characterize with a discrete wells network. 3D reflection seismology allows the extraction of additional information about the subsurface with significantly denser spatial sampling intervals. However, conventional images of the subsurface have low spatial resolution and are dominated by continuous and smooth reflections, which carry the information associated with only large-scale heterogeneities.

Diffraction images are more capable than conventional reflection images in emphasizing small-scale features associated with subsurface discontinuities. Many studies employ diffraction images as a source of additional information for interpretation. Past work has proposed an AzPWD work flow, which extends a plane-wave destruction diffraction imaging framework to account for edge-diffraction orientation and allows efficient extraction of these orientations on the basis of scanning of different azimuths.

Two modifications to the previously proposed AzPWD work flow are considered in this paper. Edge-diffraction-orientation determination is performed through a structure-tensor estimation based on PWD. A PWD-based structure tensor allows determination of edge-diffraction orientation from two volumes corresponding with PWDs applied in inline and crossline directions. Thus, no scanning over the PWD azimuth is required.

Because reflection/diffraction separation is performed on the basis of the PWD filter in the data domain, the approach pertains to a diffraction imaging framework. The structure-tensor estimation approach has been applied previously in the image space on the basis of image gradients, the highest contribution to which is connected with strong and coherent reflection events masking diffractions. The second improvement presented by the AzPWD work flow is based on additional smoothing along the edge orientation, which makes the diffraction/reflection-separation operator equivalent to the structure-oriented Sobel filter.

The AzPWD work flow is applied to an unconventional tight-gas-sand reservoir in the Cooper Basin in Western Australia. The complete paper focuses on the ­application of this work flow, which takes into account edge-diffraction phenomena and allows determination of edge-diffraction-orientation azimuth. Produced edge-diffraction orientations are consistent with the lateral distribution of major faults and channels. Moreover, several zones of “subtle” diffractivity are identified that are associated with distinct edge-diffraction-­orientation values consistent with orientations of major discontinuities.


Diffraction-imaging work flows based on PWD filters can be robust and efficient. For 3D data, PWD can be applied in either inline or crossline directions. Edge diffractors, which are associated with channel edges, faults, and the like, have lateral symmetry that aligns with the edge. Unless reflection elimination by PWD is perpendicular to the edge diffraction, the features may not be highlighted optimally.

Fig. 1 above shows a schematic of edge diffractions corresponding to a channel edge. The bottom of Fig. 1 illustrates a paleochannel in the subsurface and a scheme of seismic acquisition with zero offset (sources and receivers combined). Some of the diffraction ray paths are shown. Edge diffraction acts as a reflection along the edge and corresponds to diffraction hyperbolas in the planes perpendicular to the edge. The corresponding travel-time surface is shown in the upper part of the figure. When PWD is applied parallel to the edge, the diffraction surface will be eliminated because of its reflection-like behavior in the corresponding plane. Diffractions are highlighted optimally when PWD is applied perpendicular to the edge. For intermediate PWD directions, intermediate edge-diffraction preservation will take place. In general, edge-diffraction orientation does not align with inline and crossline directions and, therefore, reflection elimination by inline or crossline PWD may not be optimal.

The AzPWD approach uses the linearity of PWD and migration procedures and allows for efficient image estimation for arbitrary azimuthal direction. Because edge diffractions are suppressed when PWD is applied along the edge and are optimally highlighted when PWD is applied perpendicular to the edge, optimal PWD azimuth and, at the same time, edge-diffraction orientation can be determined by analyzing edge-diffraction amplitude variation under different PWD directions and picking the azimuths producing the highest absolute amplitude values for a given sample. Note that this picking can be performed for each time slice independently, which allows for target-oriented imaging and for computational cost savings.

The picking procedure corresponds to the problem of structure-orientation determination in structure-oriented smoothing work flows. Structure-tensor estimation allows avoidance of significant computation cost associated with scanning over structure orientations. Structure tensor corresponds to the smoothed outer product of the seismic amplitude gradient estimated in the image domain and allows for orientation estimation for smoothed structures. For each location of the image, structure tensor exists and the corresponding orientation of coherent structures can be determined by its eigen decomposition. The largest eigenvalue in this case corresponds to the eigenvector with orientation perpendicular to the structures.

The authors propose use of PWD filters to perform structure-tensor estimation. To make this consistent with diffraction imaging philosophy, the authors start from PWD application on the stacked and not-migrated data, migrate the corresponding inline and crossline PWD volumes, and combine them in a tensor by use of a framework.

The paper proposes implementing smoothing in the image domain along the slopes determined in the image domain and with edge preservation. Smoothed structure tensors allow ­orientation determination for smooth and robust structures.

The work flow can be summarized as follows:

  1. Perform PWD in the inline and crossline directions.
  2. Migrate volumes acquired in Step 1 using the chosen migration algorithm.
  3. Estimate structure-tensor components.
  4. Perform structure-oriented and edge-preserving smoothing of structure-tensor components.
  5. Perform eigendecomposition of the structure tensor and determine the azimuth normal to the edges.
  6. Use extracted azimuths to orient PWD filter-application direction perpendicular to the edge.

In the conventional structure-tensor approach, gradients are computed over a full-waveform image, where specular energy masks weaker diffractions associated with scattering of seismic energy on subsurface discontinuities. Diffraction imaging, in its turn, aims to boost small-scale subsurface heterogeneities. Structure-tensor estimation based on preimaging PWD volumes allows retention of information associated with diffractions. On the other hand, the previously mentioned PWD tensor estimation allows avoidance of costly scanning for structure orientation.
To take full advantage of edge-diffraction phenomena, its reflected component can be emphasized by smoothing along the edge in the direction perpendicular to the optimal PWD direction.

Unconventional Tight-Gas-Sand Reservoir

The authors have applied the AzPWD work flow to a land data set from the Cooper Basin acquired to characterize an unconventional tight-gas-sand reservoir. The target horizon corresponds to the interface between tight gas sand and coal and appears approximately at 1.72 seconds of two-way time. Here, the focus is on the result acquired with the proposed AzPWD diffraction-imaging work flow.

Diffraction-based velocity analysis is performed, and volumes corresponding to inline and crossline PWD volumes are migrated with the estimated velocity distribution. The volume describing image transformation under perturbation of PWD azimuth can be generated by using the linearity of the PWD filter and migration.

Noticeable trends corresponding to high absolute amplitude values indicate that PWD has been applied in the direction perpendicular to the edge. By picking these trends, optimal PWD azimuth distribution can be generated. Optimal PWD application direction directly corresponds to edge-diffraction orientation and is locally perpendicular to the edge.

Structure-tensor estimation is proposed as an alternative to picking for PWD filter-orientation determination. Structure tensors are formed for each image coordinate: One starts with inline and crossline PWD stacks, migrates them using the velocity distribution acquired by diffraction-based migration velocity analysis, combines them, and smooths matrix components along the structures.

Slicing through the volume with optimal PWD azimuth values results in the diffractivity distribution. It should be mentioned that generation of an image set corresponding to a range of PWD azimuths is not necessary: PWD-based structure tensor allows extraction of orientation of edge diffractions on the basis of inline and crossline PWD volumes only. Structures perpendicular to each other in orientation appear to be emphasized on the same image simultaneously. To improve the signal/noise ratio of the diffractivity image, instead of orienting the PWD filter one can use extracted azimuths to orient the Sobel filter. The Sobel filter includes smoothing along the edge and, therefore, emphasizes edge diffractions, which exhibit preferential orientation, and it suppresses noise.

Finally, by use of edge-diffraction orientation to orient the PWD filter at each location, a diffraction image can be generated. In comparison with a conventional image, subsurface discontinuities are significantly emphasized in the diffraction image.

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper URTeC 2695232, “Unconventional-Reservoir Characterization With Azimuthal Seismic Diffraction Imaging,” by Dmitrii Merzlikin, Sergey Fomel, Xinming Wu, and Mason Phillips, The University of Texas at Austin, prepared for the 2017 Unconventional Resources Technology Conference, Austin, Texas, USA, 24–26 July. The paper has not been peer reviewed.